new_motif_second_part.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SigE second part of putative binding motif\n",
    "\n",
    "- check if the high scoring motif found at centers of sigE peaks has second part\n",
    "- we will treat the motif as possible -10 or -35 part and look for conserved motif at expected distance\n",
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from Bio import SeqIO\n",
    "import os\n",
    "import shutil\n",
    "import meme_txt_parser\n",
    "import subprocess\n",
    "from IPython.display import display, Image\n",
    "from Bio import SeqUtils\n",
    "\n",
    "SC_ACCESSION = \"AL645882.2\"\n",
    "temp = 'new_motif_site'\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load genome\n",
    "sc_genome = [\n",
    "    s for s in SeqIO.parse('data/GCA_000203835.1_ASM20383v1_genomic.fna', format='fasta') if s.id == SC_ACCESSION\n",
    "][0]\n",
    "\n",
    "# load found MOTIF 1 sites"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>motif_id</th>\n",
       "      <th>sequence_name</th>\n",
       "      <th>strand</th>\n",
       "      <th>start</th>\n",
       "      <th>P-value</th>\n",
       "      <th>site</th>\n",
       "      <th>sequence_offset</th>\n",
       "      <th>stop</th>\n",
       "      <th>downstream_gene</th>\n",
       "      <th>distance_to_downstream</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_181</td>\n",
       "      <td>+</td>\n",
       "      <td>16</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>2088569</td>\n",
       "      <td>23</td>\n",
       "      <td>SCO1954</td>\n",
       "      <td>1617.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_171</td>\n",
       "      <td>+</td>\n",
       "      <td>53</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>3659392</td>\n",
       "      <td>60</td>\n",
       "      <td>SCO3310</td>\n",
       "      <td>2285.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_168</td>\n",
       "      <td>+</td>\n",
       "      <td>81</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>4074167</td>\n",
       "      <td>88</td>\n",
       "      <td>SCO3696</td>\n",
       "      <td>1683.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_156</td>\n",
       "      <td>-</td>\n",
       "      <td>43</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>6883700</td>\n",
       "      <td>50</td>\n",
       "      <td>SCO6255</td>\n",
       "      <td>5568.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_154</td>\n",
       "      <td>+</td>\n",
       "      <td>51</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>7631695</td>\n",
       "      <td>58</td>\n",
       "      <td>SCO6885</td>\n",
       "      <td>22393.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>82</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_116</td>\n",
       "      <td>-</td>\n",
       "      <td>42</td>\n",
       "      <td>0.000339</td>\n",
       "      <td>ATAGACG</td>\n",
       "      <td>3064610</td>\n",
       "      <td>49</td>\n",
       "      <td>SCO2806</td>\n",
       "      <td>88.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>83</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_178</td>\n",
       "      <td>-</td>\n",
       "      <td>53</td>\n",
       "      <td>0.000394</td>\n",
       "      <td>GTAGATG</td>\n",
       "      <td>2645931</td>\n",
       "      <td>60</td>\n",
       "      <td>SCO2458</td>\n",
       "      <td>563.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>84</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_158</td>\n",
       "      <td>+</td>\n",
       "      <td>49</td>\n",
       "      <td>0.000394</td>\n",
       "      <td>AAATAAG</td>\n",
       "      <td>6509941</td>\n",
       "      <td>56</td>\n",
       "      <td>SCO5943</td>\n",
       "      <td>28.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_187</td>\n",
       "      <td>-</td>\n",
       "      <td>36</td>\n",
       "      <td>0.000404</td>\n",
       "      <td>ATAGAAG</td>\n",
       "      <td>490356</td>\n",
       "      <td>43</td>\n",
       "      <td>SCO0468</td>\n",
       "      <td>42.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_115</td>\n",
       "      <td>+</td>\n",
       "      <td>75</td>\n",
       "      <td>0.000696</td>\n",
       "      <td>ATATAAG</td>\n",
       "      <td>3095306</td>\n",
       "      <td>82</td>\n",
       "      <td>SCO2836</td>\n",
       "      <td>118.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>87 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   motif_id sequence_name strand  start   P-value     site  sequence_offset  \\\n",
       "0    MEME-1      sigE_181      +     16  0.000046  GAAGACG          2088569   \n",
       "1    MEME-1      sigE_171      +     53  0.000046  GAAGACG          3659392   \n",
       "2    MEME-1      sigE_168      +     81  0.000046  GAAGACG          4074167   \n",
       "3    MEME-1      sigE_156      -     43  0.000046  GAAGACG          6883700   \n",
       "4    MEME-1      sigE_154      +     51  0.000046  GAAGACG          7631695   \n",
       "..      ...           ...    ...    ...       ...      ...              ...   \n",
       "82   MEME-1      sigE_116      -     42  0.000339  ATAGACG          3064610   \n",
       "83   MEME-1      sigE_178      -     53  0.000394  GTAGATG          2645931   \n",
       "84   MEME-1      sigE_158      +     49  0.000394  AAATAAG          6509941   \n",
       "85   MEME-1      sigE_187      -     36  0.000404  ATAGAAG           490356   \n",
       "86   MEME-1      sigE_115      +     75  0.000696  ATATAAG          3095306   \n",
       "\n",
       "    stop downstream_gene  distance_to_downstream  \n",
       "0     23         SCO1954                  1617.0  \n",
       "1     60         SCO3310                  2285.0  \n",
       "2     88         SCO3696                  1683.0  \n",
       "3     50         SCO6255                  5568.0  \n",
       "4     58         SCO6885                 22393.0  \n",
       "..   ...             ...                     ...  \n",
       "82    49         SCO2806                    88.0  \n",
       "83    60         SCO2458                   563.0  \n",
       "84    56         SCO5943                    28.0  \n",
       "85    43         SCO0468                    42.0  \n",
       "86    82         SCO2836                   118.0  \n",
       "\n",
       "[87 rows x 10 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m10_dir = 'peak_summit_motif_discovery'\n",
    "\n",
    "df = pd.read_excel(m10_dir + '/sigE_motif_discovery.xlsx', sheet_name='sigE-motif', skiprows=range(14))\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# prepare for motig reverse complemet\n",
    "\n",
    "df_rc = df.copy()\n",
    "df_rc.loc[df.strand == '+', 'strand'] = '-'\n",
    "df_rc.loc[df.strand == '-', 'strand'] = '+'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "os.environ[\"JUPYTER_TEMP\"] = temp\n",
    "os.environ[\"M10DIR\"] = m10_dir\n",
    "\n",
    "shutil.rmtree(temp, ignore_errors=True)\n",
    "os.makedirs(temp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# set possition relative to identified motif site\n",
    "# either a) upstream b) downstream\n",
    "\n",
    "\n",
    "# a) upstream\n",
    "\n",
    "def upstream_from_motif(genome, df, offset=15, segment_len=10):\n",
    "    upstream_segments = []\n",
    "    for r in df.itertuples():\n",
    "        motif_s = r.sequence_offset -1 + r.start\n",
    "        motif_e = r.sequence_offset -1 + r.stop\n",
    "        if r.strand == '+':\n",
    "            e = motif_s - offset\n",
    "            s = e - segment_len\n",
    "\n",
    "            segment = genome[s:e]\n",
    "\n",
    "        elif r.strand == '-':\n",
    "            s = motif_e + offset\n",
    "            e = s + segment_len\n",
    "\n",
    "            segment = genome[s:e].reverse_complement()\n",
    "\n",
    "        else:\n",
    "            raise NotImplementedError()\n",
    "\n",
    "        segment.id = r.sequence_name + '-{}_{}-{}'.format(r.strand, s, len(segment))\n",
    "        segment.description = ''\n",
    "        segment.name = ''\n",
    "        upstream_segments.append(segment)   \n",
    "    return upstream_segments\n",
    "\n",
    "\n",
    "# b) downstream\n",
    "def downstream_from_motif(genome, df, offset=15, segment_len=10):\n",
    "    downstream_segments = []\n",
    "    for r in df.itertuples():\n",
    "        motif_s = r.sequence_offset -1 + r.start\n",
    "        motif_e = r.sequence_offset -1 + r.stop\n",
    "        if r.strand == '+':\n",
    "            s = motif_e + offset\n",
    "            e = s + segment_len\n",
    "\n",
    "            segment = genome[s:e]\n",
    "\n",
    "        elif r.strand == '-':\n",
    "            e = motif_s - offset\n",
    "            s = e - segment_len\n",
    "\n",
    "            segment = genome[s:e].reverse_complement()\n",
    "\n",
    "        else:\n",
    "            raise NotImplementedError()\n",
    "\n",
    "\n",
    "        segment.id = r.sequence_name + '-{}_{}-{}'.format(r.strand, s, len(segment))\n",
    "        segment.description = ''\n",
    "        segment.name = ''\n",
    "        downstream_segments.append(segment)\n",
    "    return downstream_segments\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "upstream_segments = upstream_from_motif(sc_genome, df, 13, 14)\n",
    "downstream_segments = downstream_from_motif(sc_genome, df, 13, 14)\n",
    "\n",
    "upstream_segments_from_rc = upstream_from_motif(sc_genome, df_rc, 13, 14)\n",
    "downstream_segments_from_rc = downstream_from_motif(sc_genome, df_rc, 13, 14)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "87"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SeqIO.write(upstream_segments, os.path.join(temp, 'upstream_segments.fasta'), format='fasta')\n",
    "SeqIO.write(downstream_segments, os.path.join(temp, 'downstream_segments.fasta'), format='fasta')\n",
    "\n",
    "SeqIO.write(upstream_segments_from_rc, os.path.join(temp, 'upstream_segments_from_rc.fasta'), format='fasta')\n",
    "SeqIO.write(downstream_segments_from_rc, os.path.join(temp, 'downstream_segments_from_rc.fasta'), format='fasta')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Writing results to output directory 'new_motif_site/upstream'.\n",
      "BACKGROUND: using background model of order 1\n",
      "PRIMARY (classic): n 87 p0 87 p1 0 p2 0\n",
      "SEQUENCE GROUP USAGE-- Starts/EM: p0; Trim: p0; pvalue: p0; nsites: p0,p1,p2\n",
      "SEEDS: maxwords 1218 highwater mark: seq 87 pos 10\n",
      "BALANCE: samples 87 chars 1218 nodes 1 chars/node 1218\n",
      "Initializing the motif probability tables for 2 to 87 sites...\n",
      "nsites = 87\n",
      "Done initializing.\n",
      "all widths from min to max\n",
      "\n",
      "seqs=    87, min_w=  14, max_w=   14, total_size=     1218\n",
      "\n",
      "motif=1\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "motif=2\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "motif=3\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "Writing results to output directory 'new_motif_site/downstream'.\n",
      "BACKGROUND: using background model of order 1\n",
      "PRIMARY (classic): n 87 p0 87 p1 0 p2 0\n",
      "SEQUENCE GROUP USAGE-- Starts/EM: p0; Trim: p0; pvalue: p0; nsites: p0,p1,p2\n",
      "SEEDS: maxwords 1218 highwater mark: seq 87 pos 10\n",
      "BALANCE: samples 87 chars 1218 nodes 1 chars/node 1218\n",
      "Initializing the motif probability tables for 2 to 87 sites...\n",
      "nsites = 87\n",
      "Done initializing.\n",
      "all widths from min to max\n",
      "\n",
      "seqs=    87, min_w=  14, max_w=   14, total_size=     1218\n",
      "\n",
      "motif=1\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "motif=2\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "motif=3\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "Writing results to output directory 'new_motif_site/upstream_from_rc'.\n",
      "BACKGROUND: using background model of order 1\n",
      "PRIMARY (classic): n 87 p0 87 p1 0 p2 0\n",
      "SEQUENCE GROUP USAGE-- Starts/EM: p0; Trim: p0; pvalue: p0; nsites: p0,p1,p2\n",
      "SEEDS: maxwords 1218 highwater mark: seq 87 pos 10\n",
      "BALANCE: samples 87 chars 1218 nodes 1 chars/node 1218\n",
      "Initializing the motif probability tables for 2 to 87 sites...\n",
      "nsites = 87\n",
      "Done initializing.\n",
      "all widths from min to max\n",
      "\n",
      "seqs=    87, min_w=  14, max_w=   14, total_size=     1218\n",
      "\n",
      "motif=1\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "motif=2\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "motif=3\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "Writing results to output directory 'new_motif_site/downstream_from_rc'.\n",
      "BACKGROUND: using background model of order 1\n",
      "PRIMARY (classic): n 87 p0 87 p1 0 p2 0\n",
      "SEQUENCE GROUP USAGE-- Starts/EM: p0; Trim: p0; pvalue: p0; nsites: p0,p1,p2\n",
      "SEEDS: maxwords 1218 highwater mark: seq 87 pos 10\n",
      "BALANCE: samples 87 chars 1218 nodes 1 chars/node 1218\n",
      "Initializing the motif probability tables for 2 to 87 sites...\n",
      "nsites = 87\n",
      "Done initializing.\n",
      "all widths from min to max\n",
      "\n",
      "seqs=    87, min_w=  14, max_w=   14, total_size=     1218\n",
      "\n",
      "motif=1\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "motif=2\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n",
      "motif=3\n",
      "SEED DEPTHS: 2 4 8 16 32 64 87\n",
      "SEED WIDTHS: 4 5 6 7 8\n",
      "em: w=   8, psites=  87, iter=  40 \n"
     ]
    }
   ],
   "source": [
    "%%bash\n",
    "\n",
    "meme -oc $JUPYTER_TEMP/upstream -mod zoops -nmotifs 3 -minw 4 -maxw 8 -dna -allw -bfile $M10DIR/AL645882.2_m1.markov $JUPYTER_TEMP/upstream_segments.fasta\n",
    "meme -oc $JUPYTER_TEMP/downstream -mod zoops -nmotifs 3 -minw 4 -maxw 8 -dna -allw -bfile $M10DIR/AL645882.2_m1.markov $JUPYTER_TEMP/downstream_segments.fasta\n",
    "\n",
    "meme -oc $JUPYTER_TEMP/upstream_from_rc -mod zoops -nmotifs 3 -minw 4 -maxw 8 -dna -allw -bfile $M10DIR/AL645882.2_m1.markov $JUPYTER_TEMP/upstream_segments_from_rc.fasta\n",
    "meme -oc $JUPYTER_TEMP/downstream_from_rc -mod zoops -nmotifs 3 -minw 4 -maxw 8 -dna -allw -bfile $M10DIR/AL645882.2_m1.markov $JUPYTER_TEMP/downstream_segments_from_rc.fasta"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Check results of the motif discovery ###\n",
    "# the upstream from rc is the -35 part of Tran motif\n",
    "\n",
    "m35 = meme_txt_parser.read_file(os.path.join(temp, 'upstream_from_rc', 'meme.txt'), no_strand=True)\n",
    "m10 = meme_txt_parser.read_file(os.path.join(m10_dir, 'meme_zoops_test/meme.txt'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def _get_strand(r, retval='offset'):\n",
    "    s = r.sequence_name\n",
    "    \n",
    "    _b = s.split('_')\n",
    "    strand = _b[1][-1]\n",
    "    offset = int(_b[2].split('-')[0])\n",
    "    sl = int(_b[2].split('-')[1])\n",
    "    sequence_name = s.split('-')[0]\n",
    "        \n",
    "    return vars()[retval]\n",
    "\n",
    "m35['strand'] = m35.apply(_get_strand, retval='strand', axis=1)\n",
    "m35['offset'] = m35.apply(_get_strand, retval='offset', axis=1)\n",
    "m35['segment_length'] = m35.apply(_get_strand, retval='sl', axis=1)\n",
    "m35['sequence_name'] = m35.apply(_get_strand, retval='sequence_name', axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "ss = []\n",
    "ee = []\n",
    "for r in m35.itertuples():\n",
    "    if r.strand == '+':\n",
    "        s = r.offset+r.start-1\n",
    "        e = r.offset+r.start+len(r.site)-1\n",
    "        assert str(sc_genome.seq[s:e]) == r.site\n",
    "    elif r.strand == '-':\n",
    "        s = r.offset+(r.segment_length-len(r.site)-r.start+1)\n",
    "        e = r.offset+r.segment_length-r.start+1\n",
    "        assert str(sc_genome.seq[s:e].reverse_complement()) == r.site\n",
    "    else:\n",
    "        raise NotImplementedError()\n",
    "    ss.append(s)\n",
    "    ee.append(e)\n",
    "m35['ss'] = ss\n",
    "m35['ee'] = ee"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "m10['offset'] = m10.sequence_name.apply(lambda x: int(x.split('-')[1]))\n",
    "m10.sequence_name = m10.sequence_name.apply(lambda x: x.split('-')[0])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "ss = []\n",
    "ee = []\n",
    "for r in m10.itertuples():\n",
    "    if r.strand == '+':\n",
    "        s = r.offset+r.start-1\n",
    "        e = r.offset+r.start+len(r.site)-1\n",
    "        assert str(sc_genome.seq[s:e]) == r.site\n",
    "    elif r.strand == '-':\n",
    "        s = r.offset+r.start-1\n",
    "        e = r.offset+r.start+len(r.site)-1\n",
    "        assert str(sc_genome.seq[s:e].reverse_complement()) == r.site\n",
    "    ss.append(s)\n",
    "    ee.append(e)\n",
    "m10['ss'] = ss\n",
    "m10['ee'] = ee"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>motif_id</th>\n",
       "      <th>sequence_name</th>\n",
       "      <th>strand</th>\n",
       "      <th>start</th>\n",
       "      <th>P-value</th>\n",
       "      <th>_site_left</th>\n",
       "      <th>site</th>\n",
       "      <th>_site_right</th>\n",
       "      <th>offset</th>\n",
       "      <th>ss</th>\n",
       "      <th>ee</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_181</td>\n",
       "      <td>-</td>\n",
       "      <td>16</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>GCCCGTCCTC</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>ACCATGGAGG</td>\n",
       "      <td>2088569</td>\n",
       "      <td>2088584</td>\n",
       "      <td>2088591</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_171</td>\n",
       "      <td>-</td>\n",
       "      <td>53</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>TGCCGCCCAG</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>TCGTAGACCC</td>\n",
       "      <td>3659392</td>\n",
       "      <td>3659444</td>\n",
       "      <td>3659451</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_168</td>\n",
       "      <td>-</td>\n",
       "      <td>81</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>GCTGATCGGC</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>ACCGCGAGGC</td>\n",
       "      <td>4074167</td>\n",
       "      <td>4074247</td>\n",
       "      <td>4074254</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_156</td>\n",
       "      <td>+</td>\n",
       "      <td>43</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>GCCCCGCCAG</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>TCGAGACCGG</td>\n",
       "      <td>6883700</td>\n",
       "      <td>6883742</td>\n",
       "      <td>6883749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_154</td>\n",
       "      <td>-</td>\n",
       "      <td>51</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>CGTCGATCAA</td>\n",
       "      <td>GAAGACG</td>\n",
       "      <td>GTCATAGGCG</td>\n",
       "      <td>7631695</td>\n",
       "      <td>7631745</td>\n",
       "      <td>7631752</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  motif_id sequence_name strand  start   P-value  _site_left     site  \\\n",
       "0   MEME-1      sigE_181      -     16  0.000046  GCCCGTCCTC  GAAGACG   \n",
       "1   MEME-1      sigE_171      -     53  0.000046  TGCCGCCCAG  GAAGACG   \n",
       "2   MEME-1      sigE_168      -     81  0.000046  GCTGATCGGC  GAAGACG   \n",
       "3   MEME-1      sigE_156      +     43  0.000046  GCCCCGCCAG  GAAGACG   \n",
       "4   MEME-1      sigE_154      -     51  0.000046  CGTCGATCAA  GAAGACG   \n",
       "\n",
       "  _site_right   offset       ss       ee  \n",
       "0  ACCATGGAGG  2088569  2088584  2088591  \n",
       "1  TCGTAGACCC  3659392  3659444  3659451  \n",
       "2  ACCGCGAGGC  4074167  4074247  4074254  \n",
       "3  TCGAGACCGG  6883700  6883742  6883749  \n",
       "4  GTCATAGGCG  7631695  7631745  7631752  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "strand_invert = {'-': '+', '+': '-'}\n",
    "m10.strand = m10.strand.apply(lambda x: strand_invert[x])\n",
    "\n",
    "m10.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>motif_id</th>\n",
       "      <th>sequence_name</th>\n",
       "      <th>start</th>\n",
       "      <th>P-value</th>\n",
       "      <th>_site_left</th>\n",
       "      <th>site</th>\n",
       "      <th>_site_right</th>\n",
       "      <th>strand</th>\n",
       "      <th>offset</th>\n",
       "      <th>segment_length</th>\n",
       "      <th>ss</th>\n",
       "      <th>ee</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_129</td>\n",
       "      <td>7</td>\n",
       "      <td>0.000032</td>\n",
       "      <td>CCGGGC</td>\n",
       "      <td>CAACGTTT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>+</td>\n",
       "      <td>2502720</td>\n",
       "      <td>14</td>\n",
       "      <td>2502726</td>\n",
       "      <td>2502734</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_154</td>\n",
       "      <td>5</td>\n",
       "      <td>0.000047</td>\n",
       "      <td>ATGG</td>\n",
       "      <td>GAACCGTA</td>\n",
       "      <td>TC</td>\n",
       "      <td>-</td>\n",
       "      <td>7631765</td>\n",
       "      <td>14</td>\n",
       "      <td>7631767</td>\n",
       "      <td>7631775</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_10</td>\n",
       "      <td>7</td>\n",
       "      <td>0.000077</td>\n",
       "      <td>GCTGTG</td>\n",
       "      <td>GAACGTCA</td>\n",
       "      <td>NaN</td>\n",
       "      <td>+</td>\n",
       "      <td>7565221</td>\n",
       "      <td>14</td>\n",
       "      <td>7565227</td>\n",
       "      <td>7565235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_26</td>\n",
       "      <td>1</td>\n",
       "      <td>0.000077</td>\n",
       "      <td>.</td>\n",
       "      <td>GAACGTCA</td>\n",
       "      <td>ACTTCT</td>\n",
       "      <td>+</td>\n",
       "      <td>6720644</td>\n",
       "      <td>14</td>\n",
       "      <td>6720644</td>\n",
       "      <td>6720652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>MEME-1</td>\n",
       "      <td>sigE_21</td>\n",
       "      <td>7</td>\n",
       "      <td>0.000104</td>\n",
       "      <td>GCGGGC</td>\n",
       "      <td>GAACGGTT</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-</td>\n",
       "      <td>7016731</td>\n",
       "      <td>14</td>\n",
       "      <td>7016731</td>\n",
       "      <td>7016739</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  motif_id sequence_name  start   P-value _site_left      site _site_right  \\\n",
       "0   MEME-1      sigE_129      7  0.000032     CCGGGC  CAACGTTT         NaN   \n",
       "1   MEME-1      sigE_154      5  0.000047       ATGG  GAACCGTA          TC   \n",
       "2   MEME-1       sigE_10      7  0.000077     GCTGTG  GAACGTCA         NaN   \n",
       "3   MEME-1       sigE_26      1  0.000077          .  GAACGTCA      ACTTCT   \n",
       "4   MEME-1       sigE_21      7  0.000104     GCGGGC  GAACGGTT         NaN   \n",
       "\n",
       "  strand   offset  segment_length       ss       ee  \n",
       "0      +  2502720              14  2502726  2502734  \n",
       "1      -  7631765              14  7631767  7631775  \n",
       "2      +  7565221              14  7565227  7565235  \n",
       "3      +  6720644              14  6720644  6720652  \n",
       "4      -  7016731              14  7016731  7016739  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m35.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "_m10 = m10.loc[m10.motif_id == 'MEME-1', ['sequence_name', 'strand', 'ss', 'ee']]\n",
    "_m35 = m35.loc[m35.motif_id == 'MEME-1', ['sequence_name', 'strand', 'ss', 'ee']]\n",
    "\n",
    "mm = pd.merge(_m35, _m10, on=['sequence_name', 'strand'], suffixes=['_m35', '_m10'])\n",
    "\n",
    "dd = []\n",
    "for r in mm.itertuples():\n",
    "    if r.strand == '+':\n",
    "        d = r.ss_m10 - r.ee_m35\n",
    "    elif r.strand == '-':\n",
    "        d = r.ss_m35 - r.ee_m10\n",
    "    else:\n",
    "        raise NotImplementedError()\n",
    "    dd.append(d) \n",
    "mm['dist'] = dd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CAACGTTTCCGTTCGAATTAT------CGTCTTC\n",
      "GAACCGTATCTGTCGCCTATGAC----CGTCTTC\n",
      "GAACGTCATAGCCGTCTGGTG------CGTCTAC\n",
      "GAACGTCAACTTCTTCAACCCGGACGCCGTCTAC\n",
      "GAACGGTTTCCTCTCCCCCGT------CATCTTC\n",
      "CAACCGTTCGATGCCCCGGAA------CGTCTTC\n",
      "CAACGTCTCGCGCGCCTACGG------CGTCTTC\n",
      "GAACCCTTCGCTTGTCCCTGTG-----CGTCTTG\n",
      "CAACGGTACGCCCCGGTTACTGGA---CCTCTTC\n",
      "CAACCGCACGTACGCGGAGAACGT---CTTCTTC\n",
      "CAACATCTCCTCCTTCAACAC------CGTCTTC\n",
      "GAACGACTACCTGGTCGCCAC------CGTCTTC\n",
      "GAACGTTCCTCACTCCGCCAT------CGTCTAC\n",
      "CAACGGCACGCTGCAGACCGA------CATCTTC\n",
      "CAACCTCGCGCAGCCCCTCAC------CGTCTTG\n",
      "GAACCAGTAGGTATGTCGTTCT-----CGTCTTC\n",
      "GAACCGAACCGGTTCCCGTGG------CGTATTC\n",
      "CAACCCGAGGGGCGATGCCCG------CGTCTAT\n",
      "GAACCCTCAGCCTCCTCAGAC------CCTCTTC\n",
      "CAACCTGGTCCCCGTTTTCGT------CGTCTTC\n",
      "GAACGGCGAGCGCGGCCGCACCGT---CTTCTTC\n",
      "CAACCACGGGACCGGGTCGAG------CGTCTTT\n",
      "GAACGACGTGCAGGCGGCGGA------CGTCTTC\n",
      "GAACGCAACCCGGCGGTCTACGATGC-CATCTTC\n",
      "CAACCTGCACGGGGCGTACGA------CTTCTTC\n",
      "CAACCCTCAGGCGGTACGGGC------CGTCTTC\n",
      "GAAGCTCTCATCTCGTTGTCAG-----CTTCTTT\n",
      "GAACCGGGAGACCTGCTCGAT------CGTCTTC\n",
      "GAACACGACGTTCGGGGAGAA------CGTCTTC\n",
      "GAAGCGTTCCTTCCTCGTGAT------CGTCTTC\n",
      "GAACCGGCTCCGGGTCCTCGA------CGTCTTT\n",
      "CAACACCGCGTATCGCCACATGAC---CGTCTTC\n"
     ]
    }
   ],
   "source": [
    "# extract motif sites\n",
    "# check thi sites\n",
    "fw = str(sc_genome.seq)\n",
    "rc = str(sc_genome.seq.reverse_complement())\n",
    "promoters = []\n",
    "max_dist = mm.dist.max()\n",
    "for r in mm.itertuples():\n",
    "    \n",
    "    if r.strand == '+':\n",
    "        promoter = str(sc_genome.seq[r.ss_m35:r.ee_m35]) + str(sc_genome.seq[r.ee_m35:r.ss_m10]) + '-'*(max_dist-r.dist) + str(sc_genome.seq[r.ss_m10:r.ee_m10])\n",
    "    elif r.strand == '-':\n",
    "        promoter = ''.join([\n",
    "            str(sc_genome.seq[r.ss_m35:r.ee_m35].reverse_complement()),\n",
    "            str(sc_genome.seq[r.ee_m10:r.ss_m35].reverse_complement()) + '-'*(max_dist-r.dist),\n",
    "            str(sc_genome.seq[r.ss_m10:r.ee_m10].reverse_complement())\n",
    "        ])\n",
    "    else:\n",
    "        raise NotImplementedError()\n",
    "    \n",
    "    promoters.append(promoter)\n",
    "    \n",
    "    if promoter.replace('-', '') in fw:\n",
    "        pass\n",
    "    elif promoter.replace('-', '') in rc:\n",
    "        pass\n",
    "    else:\n",
    "        AssertionError('not found')\n",
    "    print(promoter)\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# use only spacers 16-19 counting between AACC and TCTTC\n",
    "\n",
    "motif_dir = 'high_scoring_sites'\n",
    "os.makedirs(os.path.join(temp, motif_dir), exist_ok=True)\n",
    "\n",
    "p_variants = [\n",
    "    (22, 27, '19'),\n",
    "    (21, 27, '18'),\n",
    "    (20, 27, '17'),\n",
    "    (19, 27, '16')\n",
    "]\n",
    "for p in p_variants:\n",
    "    pms = [s[:p[0]] + s[p[1]:] for s in promoters]\n",
    "    \n",
    "    with open(os.path.join(temp, motif_dir, 'hsm_{}.txt'.format(p[2])), 'w') as t:\n",
    "        t.write('\\n'.join(pms) + '\\n')\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "motifs = ['hsm_{}'.format(p[2]) for p in p_variants]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SeqIO.write([sc_genome], os.path.join(temp, 'AL645882.2.fasta'), format='fasta')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[KConverted 4 motifs.\n",
      "0 conversion errors.\n"
     ]
    }
   ],
   "source": [
    "%%bash\n",
    "fasta-get-markov -m 1 $JUPYTER_TEMP/AL645882.2.fasta > $JUPYTER_TEMP/AL645882.2_m1.markov\n",
    "sites2meme -bg $JUPYTER_TEMP/AL645882.2_m1.markov $JUPYTER_TEMP/high_scoring_sites > $JUPYTER_TEMP/high_scoring_sites.meme"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "hsm_19\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {
      "image/png": {
       "width": 500
      }
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "hsm_18\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {
      "image/png": {
       "width": 500
      }
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "hsm_17\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {
      "image/png": {
       "width": 500
      }
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "hsm_16\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {
      "image/png": {
       "width": 500
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for motif in motifs:\n",
    "    r = subprocess.call(\n",
    "        ['ceqlogo', \n",
    "         '-i', os.path.join(temp, 'high_scoring_sites.meme'), \n",
    "         '-m', motif,\n",
    "         '-f', 'PNG',\n",
    "         '-o', os.path.join(temp, motif + '.png')\n",
    "        ]\n",
    "    )\n",
    "    if r:\n",
    "        raise ChildProcessError\n",
    "        \n",
    "    r = subprocess.call(\n",
    "        ['ceqlogo', \n",
    "         '-i', os.path.join(temp, 'high_scoring_sites.meme'), \n",
    "         '-m', motif,\n",
    "         '-f', 'EPS',\n",
    "         '-o', os.path.join(temp, motif + '.eps')\n",
    "        ]\n",
    "    )\n",
    "    if r:\n",
    "        raise ChildProcessError\n",
    "        \n",
    "# todo show all motifs\n",
    "for motif in motifs:\n",
    "    print(motif)\n",
    "    display(Image(os.path.join(temp, motif + '.png'), width=500))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>start</th>\n",
       "      <th>end</th>\n",
       "      <th>length</th>\n",
       "      <th>abs_summit</th>\n",
       "      <th>pileup</th>\n",
       "      <th>neglog10pvalue</th>\n",
       "      <th>fold_enrichment</th>\n",
       "      <th>neglog10qvalue</th>\n",
       "      <th>norm/EtOH</th>\n",
       "      <th>sequence_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>8662666</td>\n",
       "      <td>8663025</td>\n",
       "      <td>360</td>\n",
       "      <td>8662837</td>\n",
       "      <td>4159.26</td>\n",
       "      <td>399.36166</td>\n",
       "      <td>2.10648</td>\n",
       "      <td>397.12149</td>\n",
       "      <td>1</td>\n",
       "      <td>sigE_0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>8483428</td>\n",
       "      <td>8484106</td>\n",
       "      <td>679</td>\n",
       "      <td>8483707</td>\n",
       "      <td>9107.42</td>\n",
       "      <td>2952.46606</td>\n",
       "      <td>4.61191</td>\n",
       "      <td>2948.32202</td>\n",
       "      <td>1</td>\n",
       "      <td>sigE_1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>8038581</td>\n",
       "      <td>8038938</td>\n",
       "      <td>358</td>\n",
       "      <td>8038755</td>\n",
       "      <td>6019.20</td>\n",
       "      <td>1160.16821</td>\n",
       "      <td>3.04824</td>\n",
       "      <td>1157.51782</td>\n",
       "      <td>1</td>\n",
       "      <td>sigE_2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>8005927</td>\n",
       "      <td>8006952</td>\n",
       "      <td>1026</td>\n",
       "      <td>8006657</td>\n",
       "      <td>9276.34</td>\n",
       "      <td>1427.76892</td>\n",
       "      <td>2.65750</td>\n",
       "      <td>1425.00598</td>\n",
       "      <td>1</td>\n",
       "      <td>sigE_3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7885747</td>\n",
       "      <td>7886423</td>\n",
       "      <td>677</td>\n",
       "      <td>7886038</td>\n",
       "      <td>8795.63</td>\n",
       "      <td>2617.73657</td>\n",
       "      <td>4.26193</td>\n",
       "      <td>2614.10645</td>\n",
       "      <td>1</td>\n",
       "      <td>sigE_4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>188</th>\n",
       "      <td>6107752</td>\n",
       "      <td>6107920</td>\n",
       "      <td>169</td>\n",
       "      <td>6107856</td>\n",
       "      <td>7593.41</td>\n",
       "      <td>1068.68542</td>\n",
       "      <td>2.52894</td>\n",
       "      <td>1066.07446</td>\n",
       "      <td>1</td>\n",
       "      <td>sigE_188</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>189</th>\n",
       "      <td>5999380</td>\n",
       "      <td>5999532</td>\n",
       "      <td>153</td>\n",
       "      <td>5999460</td>\n",
       "      <td>6709.26</td>\n",
       "      <td>833.43805</td>\n",
       "      <td>2.37112</td>\n",
       "      <td>830.94513</td>\n",
       "      <td>x</td>\n",
       "      <td>sigE_189</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>190</th>\n",
       "      <td>4336288</td>\n",
       "      <td>4336472</td>\n",
       "      <td>185</td>\n",
       "      <td>4336360</td>\n",
       "      <td>7307.68</td>\n",
       "      <td>841.24341</td>\n",
       "      <td>2.28611</td>\n",
       "      <td>838.74652</td>\n",
       "      <td>1</td>\n",
       "      <td>sigE_190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>191</th>\n",
       "      <td>1869170</td>\n",
       "      <td>1869322</td>\n",
       "      <td>153</td>\n",
       "      <td>1869226</td>\n",
       "      <td>9659.11</td>\n",
       "      <td>3326.14575</td>\n",
       "      <td>4.89125</td>\n",
       "      <td>3321.38403</td>\n",
       "      <td>1</td>\n",
       "      <td>sigE_191</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>192</th>\n",
       "      <td>3958080</td>\n",
       "      <td>3958352</td>\n",
       "      <td>273</td>\n",
       "      <td>3958160</td>\n",
       "      <td>10132.63</td>\n",
       "      <td>2805.60889</td>\n",
       "      <td>4.00254</td>\n",
       "      <td>2801.70435</td>\n",
       "      <td>1</td>\n",
       "      <td>sigE_192</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>193 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       start      end  length  abs_summit    pileup  neglog10pvalue  \\\n",
       "0    8662666  8663025     360     8662837   4159.26       399.36166   \n",
       "1    8483428  8484106     679     8483707   9107.42      2952.46606   \n",
       "2    8038581  8038938     358     8038755   6019.20      1160.16821   \n",
       "3    8005927  8006952    1026     8006657   9276.34      1427.76892   \n",
       "4    7885747  7886423     677     7886038   8795.63      2617.73657   \n",
       "..       ...      ...     ...         ...       ...             ...   \n",
       "188  6107752  6107920     169     6107856   7593.41      1068.68542   \n",
       "189  5999380  5999532     153     5999460   6709.26       833.43805   \n",
       "190  4336288  4336472     185     4336360   7307.68       841.24341   \n",
       "191  1869170  1869322     153     1869226   9659.11      3326.14575   \n",
       "192  3958080  3958352     273     3958160  10132.63      2805.60889   \n",
       "\n",
       "     fold_enrichment  neglog10qvalue norm/EtOH sequence_name  \n",
       "0            2.10648       397.12149         1        sigE_0  \n",
       "1            4.61191      2948.32202         1        sigE_1  \n",
       "2            3.04824      1157.51782         1        sigE_2  \n",
       "3            2.65750      1425.00598         1        sigE_3  \n",
       "4            4.26193      2614.10645         1        sigE_4  \n",
       "..               ...             ...       ...           ...  \n",
       "188          2.52894      1066.07446         1      sigE_188  \n",
       "189          2.37112       830.94513         x      sigE_189  \n",
       "190          2.28611       838.74652         1      sigE_190  \n",
       "191          4.89125      3321.38403         1      sigE_191  \n",
       "192          4.00254      2801.70435         1      sigE_192  \n",
       "\n",
       "[193 rows x 10 columns]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load the SigE peaks\n",
    "df_e = pd.read_excel('data/sige_macs_peaks_curated.xlsx', sheet_name='sigE_peaks')\n",
    "df_e"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# process the dataframe\n",
    "sigE_summits = {r.sequence_name: r.abs_summit for r in df_e.itertuples()}\n",
    "\n",
    "peak_seqs2 = []\n",
    "for id, s, e in df_e.loc[:, ['sequence_name', 'start', 'end']].values:\n",
    "    peak_seqs2.append(\n",
    "        '>{}-{}-{}\\n{}\\n'.format(id, s, e-s+1, str(sc_genome.seq[s:e]))\n",
    "   )\n",
    "with open(os.path.join(temp, 'sigE_peaks_whole.fasta'), 'w') as o:\n",
    "    o.write('\\n'.join(peak_seqs2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using motif +hsm_16 of width 26.\n",
      "Using motif -hsm_16 of width 26.\n",
      "Computing q-values.\n",
      "  Estimating pi_0 from a uniformly sampled set of 10000 p-values.\n",
      "  Estimating pi_0.\n",
      "  Estimated pi_0=0.984138\n",
      "Using motif +hsm_17 of width 27.\n",
      "Using motif -hsm_17 of width 27.\n",
      "Computing q-values.\n",
      "  Estimating pi_0 from a uniformly sampled set of 10000 p-values.\n",
      "  Estimating pi_0.\n",
      "  Estimated pi_0=0.981573\n",
      "Using motif +hsm_18 of width 28.\n",
      "Using motif -hsm_18 of width 28.\n",
      "Computing q-values.\n",
      "  Estimating pi_0 from a uniformly sampled set of 10000 p-values.\n",
      "  Estimating pi_0.\n",
      "  Estimated pi_0=0.934419\n",
      "Using motif +hsm_19 of width 29.\n",
      "Using motif -hsm_19 of width 29.\n",
      "Computing q-values.\n",
      "  Estimating pi_0 from a uniformly sampled set of 10000 p-values.\n",
      "  Estimating pi_0.\n",
      "  Estimated pi_0=0.99\n"
     ]
    }
   ],
   "source": [
    "%%bash\n",
    "fimo --bfile $JUPYTER_TEMP/AL645882.2_m1.markov --oc $JUPYTER_TEMP/sigE_peaks_whole --thresh 0.05 $JUPYTER_TEMP/high_scoring_sites.meme $JUPYTER_TEMP/sigE_peaks_whole.fasta"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "fimo_motif = pd.read_table(os.path.join(temp, 'sigE_peaks_whole', 'fimo.tsv'), comment='#')\n",
    "fimo_motif['offset'] = fimo_motif.sequence_name.apply(lambda x: int(x.split('-')[1]))\n",
    "fimo_motif['sequence_name'] = fimo_motif.sequence_name.apply(lambda x: x.split('-')[0])\n",
    "\n",
    "fimo_motif_wf = fimo_motif.copy()\n",
    "# skip filtering as JV wants to analyze the spacer properties\n",
    "\n",
    "# basic filtering\n",
    "# dropping simple duplicates while keeping hit with better p-val\n",
    "_m = []\n",
    "for (sname, strand), nn in fimo_motif.groupby(['sequence_name', 'strand']):\n",
    "    rr = nn.sort_values('p-value').drop_duplicates('start').drop_duplicates('stop').copy()\n",
    "    if strand == '+':\n",
    "        rr['dist2abs_summit'] = abs((rr.start + rr.offset) - sigE_summits[sname])\n",
    "    elif strand == '-':\n",
    "        rr['dist2abs_summit'] = abs((rr.stop + rr.offset) - sigE_summits[sname])\n",
    "    _m.append(rr)\n",
    "fimo_motif = pd.concat(_m, axis=0).reset_index()\n",
    "\n",
    "# but we still need to compute the distance 2 absolute summit\n",
    "fimo_motif_wf['dist2abs_summit'] = float('NaN')\n",
    "fip = fimo_motif_wf.loc[fimo_motif_wf.strand == '+']\n",
    "fim = fimo_motif_wf.loc[fimo_motif_wf.strand == '-']\n",
    "fimo_motif_wf.loc[fimo_motif_wf.strand == '+', 'dist2abs_summit'] = abs((fip.start + fip.offset) - [sigE_summits[i] for i in fip.sequence_name.values])\n",
    "fimo_motif_wf.loc[fimo_motif_wf.strand == '-', 'dist2abs_summit'] = abs((fim.stop + fim.offset) - [sigE_summits[i] for i in fim.sequence_name.values])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>seqid</th>\n",
       "      <th>source</th>\n",
       "      <th>type</th>\n",
       "      <th>start</th>\n",
       "      <th>end</th>\n",
       "      <th>score</th>\n",
       "      <th>strand</th>\n",
       "      <th>phase</th>\n",
       "      <th>Note</th>\n",
       "      <th>Name</th>\n",
       "      <th>gene</th>\n",
       "      <th>gene_synonym</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>AL645882.2</td>\n",
       "      <td>EMBL</td>\n",
       "      <td>region</td>\n",
       "      <td>1</td>\n",
       "      <td>8667507</td>\n",
       "      <td>.</td>\n",
       "      <td>+</td>\n",
       "      <td>.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>AL645882.2</td>\n",
       "      <td>EMBL</td>\n",
       "      <td>sequence_feature</td>\n",
       "      <td>1</td>\n",
       "      <td>21653</td>\n",
       "      <td>.</td>\n",
       "      <td>+</td>\n",
       "      <td>.</td>\n",
       "      <td>TIR-L. Left hand chromosome end terminal inver...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>AL645882.2</td>\n",
       "      <td>EMBL</td>\n",
       "      <td>ribosome_entry_site</td>\n",
       "      <td>435</td>\n",
       "      <td>440</td>\n",
       "      <td>.</td>\n",
       "      <td>+</td>\n",
       "      <td>.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>AL645882.2</td>\n",
       "      <td>EMBL</td>\n",
       "      <td>gene</td>\n",
       "      <td>446</td>\n",
       "      <td>1123</td>\n",
       "      <td>.</td>\n",
       "      <td>+</td>\n",
       "      <td>.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>SCO0001</td>\n",
       "      <td>SCO0001</td>\n",
       "      <td>SCEND.02c</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>AL645882.2</td>\n",
       "      <td>EMBL</td>\n",
       "      <td>CDS</td>\n",
       "      <td>446</td>\n",
       "      <td>1123</td>\n",
       "      <td>.</td>\n",
       "      <td>+</td>\n",
       "      <td>0</td>\n",
       "      <td>SCEND.02c%2C unknown%2C doubtful CDS%2C len: 2...</td>\n",
       "      <td>CAD30875.1</td>\n",
       "      <td>SCO0001</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        seqid source                 type  start      end score strand phase  \\\n",
       "0  AL645882.2   EMBL               region      1  8667507     .      +     .   \n",
       "1  AL645882.2   EMBL     sequence_feature      1    21653     .      +     .   \n",
       "2  AL645882.2   EMBL  ribosome_entry_site    435      440     .      +     .   \n",
       "3  AL645882.2   EMBL                 gene    446     1123     .      +     .   \n",
       "4  AL645882.2   EMBL                  CDS    446     1123     .      +     0   \n",
       "\n",
       "                                                Note        Name     gene  \\\n",
       "0                                                NaN         NaN      NaN   \n",
       "1  TIR-L. Left hand chromosome end terminal inver...         NaN      NaN   \n",
       "2                                                NaN         NaN      NaN   \n",
       "3                                                NaN     SCO0001  SCO0001   \n",
       "4  SCEND.02c%2C unknown%2C doubtful CDS%2C len: 2...  CAD30875.1  SCO0001   \n",
       "\n",
       "  gene_synonym  \n",
       "0          NaN  \n",
       "1          NaN  \n",
       "2          NaN  \n",
       "3    SCEND.02c  \n",
       "4          NaN  "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features = pd.read_table(\n",
    "    'data/GCA_000203835.1_ASM20383v1_genomic.gff', comment=\"#\", names=['seqid', 'source', 'type', 'start', 'end', 'score', 'strand', 'phase', 'attributes']\n",
    ")\n",
    "\n",
    "features = features.loc[features.seqid == SC_ACCESSION]\n",
    "\n",
    "def gff_attributes_mapper(gff_attr):\n",
    "    d = dict()\n",
    "    for attr in gff_attr.split(';'):\n",
    "        k,v = attr.split('=')\n",
    "        d[k] = v\n",
    "    return d\n",
    "\n",
    "features = pd.concat([features, pd.DataFrame(map(lambda x:x, features.attributes.apply(gff_attributes_mapper)))], axis=1)\n",
    "features = features.drop(columns=[\n",
    "    'attributes', 'Dbxref', 'product', 'protein_id', 'transl_table', 'partial', 'gene_biotype', 'Parent', 'start_range',\n",
    "    'regulatory_class', 'end_range', 'serovar', 'is_ordered', 'part', 'ID','mol_type', 'gbkey', 'old-name', 'bound_moiety', 'pseudo'\n",
    "])\n",
    "features.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "def downstream_from_fimo(features, df, types=['gene', 'CDS', 'pseudogene']):\n",
    "    nan = float('NaN')\n",
    "    empty_rec = pd.Series({\n",
    "        'seqid': nan,\n",
    "        'source': nan,\n",
    "        'type': nan,\n",
    "        'start': nan,\n",
    "        'end': nan,\n",
    "        'score': nan,\n",
    "        'strand': nan,\n",
    "        'phase': nan,\n",
    "        'Note': nan,\n",
    "        'Name': nan,\n",
    "        'gene': nan,\n",
    "        'gene_synonym': nan,\n",
    "        'dist2downstream': nan\n",
    "    })\n",
    "    \n",
    "    features = features.copy()\n",
    "    features['dist2downstream'] = 10*6    # some high value\n",
    "    first_downstream = []\n",
    "    for r in df.itertuples():       \n",
    "        if r.strand == '+':\n",
    "            sel = features.loc[(features.strand == r.strand) & (features.start > r.stop + r.offset) & (features.type.isin(types)), :]\n",
    "            if len(sel) == 0:\n",
    "                first_downstream.append(empty_rec)\n",
    "                continue\n",
    "            fr = sel.loc[sel.start.idxmin()].copy()\n",
    "            fr['dist2downstream'] = fr.start - (r.stop + r.offset) - 1\n",
    "        elif r.strand == '-':\n",
    "            sel = features.loc[(features.strand == r.strand) & (features.end < r.start + r.offset) & (features.type.isin(types)), :]\n",
    "            if len(sel) == 0:\n",
    "                first_downstream.append(empty_rec)\n",
    "                continue\n",
    "            fr = sel.loc[sel.end.idxmax()].copy()\n",
    "            fr['dist2downstream'] = (r.start + r.offset) - fr.end - 1\n",
    "        else:\n",
    "            raise NotImplementedError()\n",
    "        \n",
    "        first_downstream.append(fr)\n",
    "    return pd.DataFrame().from_records(first_downstream)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "first_downstream = downstream_from_fimo(features, fimo_motif)\n",
    "first_downstream_wf = downstream_from_fimo(features, fimo_motif_wf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "fimo_motifs = pd.concat([fimo_motif, first_downstream.loc[:, ['gene', 'dist2downstream']]], axis=1)\n",
    "fimo_motifs_wf = pd.concat([fimo_motif_wf, first_downstream_wf.loc[:, ['gene', 'dist2downstream']]], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using motif +HrdB-10 of width 10.\n",
      "Using motif -HrdB-10 of width 10.\n",
      "Computing q-values.\n",
      "  Estimating pi_0 from a uniformly sampled set of 10000 p-values.\n",
      "  Estimating pi_0.\n",
      "  Estimated pi_0=0.8918\n"
     ]
    }
   ],
   "source": [
    "%%bash\n",
    "fimo --bfile $JUPYTER_TEMP/AL645882.2_m1.markov --oc $JUPYTER_TEMP/fimo_hrdb_output --thresh 0.05 data/motifs/hrdb_-10_tr.meme $JUPYTER_TEMP/sigE_peaks_whole.fasta"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "# update\n",
    "# select the HrdB sites as follows\n",
    "# find the best one for each peak and strand and check if it is \"Exact\"\n",
    "\n",
    "hrdb_sites = pd.read_table(os.path.join(temp, 'fimo_hrdb_output/fimo.tsv'), comment=\"#\")\n",
    "hrdb_sites = hrdb_sites.drop('motif_alt_id', axis=1)\n",
    "hrdb_sites['offset'] = hrdb_sites.sequence_name.str.split('-').str[1].astype(int)\n",
    "hrdb_sites.sequence_name = hrdb_sites.sequence_name.str.split('-').str[0]\n",
    "hrdb_sites.head()\n",
    "\n",
    "hrdb_sites['exact'] = False\n",
    "hrdb_sites['best'] = False\n",
    "hrdb_sites.loc[hrdb_sites.matched_sequence.apply(lambda x: (x[2] == \"T\") and (x[3] == \"A\") and (x[7] == \"T\")), 'exact'] = True\n",
    "ex_hrdb_sites = hrdb_sites.loc[hrdb_sites.exact]\n",
    "\n",
    "hrdb_sites.loc[hrdb_sites.groupby(['sequence_name', 'strand'])['p-value'].idxmin(), 'best'] = True\n",
    "b_and_p_hrdb_sites = hrdb_sites.loc[hrdb_sites.exact & hrdb_sites.best].reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>motif_id</th>\n",
       "      <th>sequence_name</th>\n",
       "      <th>start</th>\n",
       "      <th>stop</th>\n",
       "      <th>strand</th>\n",
       "      <th>score</th>\n",
       "      <th>p-value</th>\n",
       "      <th>q-value</th>\n",
       "      <th>matched_sequence</th>\n",
       "      <th>offset</th>\n",
       "      <th>exact</th>\n",
       "      <th>best</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>HrdB-10</td>\n",
       "      <td>sigE_3</td>\n",
       "      <td>261</td>\n",
       "      <td>270</td>\n",
       "      <td>-</td>\n",
       "      <td>10.7797</td>\n",
       "      <td>0.000011</td>\n",
       "      <td>0.829</td>\n",
       "      <td>GGTACGATGG</td>\n",
       "      <td>8005927</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>HrdB-10</td>\n",
       "      <td>sigE_183</td>\n",
       "      <td>262</td>\n",
       "      <td>271</td>\n",
       "      <td>-</td>\n",
       "      <td>10.5254</td>\n",
       "      <td>0.000019</td>\n",
       "      <td>0.829</td>\n",
       "      <td>GTTAGACTGG</td>\n",
       "      <td>1969219</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>HrdB-10</td>\n",
       "      <td>sigE_58</td>\n",
       "      <td>295</td>\n",
       "      <td>304</td>\n",
       "      <td>-</td>\n",
       "      <td>10.4576</td>\n",
       "      <td>0.000021</td>\n",
       "      <td>0.829</td>\n",
       "      <td>GGTAAAGTGA</td>\n",
       "      <td>5474350</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>HrdB-10</td>\n",
       "      <td>sigE_85</td>\n",
       "      <td>946</td>\n",
       "      <td>955</td>\n",
       "      <td>-</td>\n",
       "      <td>10.2542</td>\n",
       "      <td>0.000039</td>\n",
       "      <td>0.829</td>\n",
       "      <td>GGTACCATGA</td>\n",
       "      <td>4325832</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>HrdB-10</td>\n",
       "      <td>sigE_172</td>\n",
       "      <td>821</td>\n",
       "      <td>830</td>\n",
       "      <td>-</td>\n",
       "      <td>10.1695</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>0.829</td>\n",
       "      <td>GGTATAGTTG</td>\n",
       "      <td>3645245</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   index motif_id sequence_name  start  stop strand    score   p-value  \\\n",
       "0      0  HrdB-10        sigE_3    261   270      -  10.7797  0.000011   \n",
       "1      1  HrdB-10      sigE_183    262   271      -  10.5254  0.000019   \n",
       "2      2  HrdB-10       sigE_58    295   304      -  10.4576  0.000021   \n",
       "3      3  HrdB-10       sigE_85    946   955      -  10.2542  0.000039   \n",
       "4      4  HrdB-10      sigE_172    821   830      -  10.1695  0.000042   \n",
       "\n",
       "   q-value matched_sequence   offset  exact  best  \n",
       "0    0.829       GGTACGATGG  8005927   True  True  \n",
       "1    0.829       GTTAGACTGG  1969219   True  True  \n",
       "2    0.829       GGTAAAGTGA  5474350   True  True  \n",
       "3    0.829       GGTACCATGA  4325832   True  True  \n",
       "4    0.829       GGTATAGTTG  3645245   True  True  "
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "b_and_p_hrdb_sites.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "hrdb_downstream_features = downstream_from_fimo(features, b_and_p_hrdb_sites)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>motif_id</th>\n",
       "      <th>sequence_name</th>\n",
       "      <th>start</th>\n",
       "      <th>stop</th>\n",
       "      <th>strand</th>\n",
       "      <th>score</th>\n",
       "      <th>p-value</th>\n",
       "      <th>q-value</th>\n",
       "      <th>matched_sequence</th>\n",
       "      <th>offset</th>\n",
       "      <th>exact</th>\n",
       "      <th>best</th>\n",
       "      <th>gene</th>\n",
       "      <th>dist2downstream</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>HrdB-10</td>\n",
       "      <td>sigE_3</td>\n",
       "      <td>261</td>\n",
       "      <td>270</td>\n",
       "      <td>-</td>\n",
       "      <td>10.7797</td>\n",
       "      <td>0.000011</td>\n",
       "      <td>0.829</td>\n",
       "      <td>GGTACGATGG</td>\n",
       "      <td>8005927</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>SCO7200</td>\n",
       "      <td>2071</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>HrdB-10</td>\n",
       "      <td>sigE_183</td>\n",
       "      <td>262</td>\n",
       "      <td>271</td>\n",
       "      <td>-</td>\n",
       "      <td>10.5254</td>\n",
       "      <td>0.000019</td>\n",
       "      <td>0.829</td>\n",
       "      <td>GTTAGACTGG</td>\n",
       "      <td>1969219</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>SCO1840</td>\n",
       "      <td>88</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   index motif_id sequence_name  start  stop strand    score   p-value  \\\n",
       "0      0  HrdB-10        sigE_3    261   270      -  10.7797  0.000011   \n",
       "1      1  HrdB-10      sigE_183    262   271      -  10.5254  0.000019   \n",
       "\n",
       "   q-value matched_sequence   offset  exact  best     gene  dist2downstream  \n",
       "0    0.829       GGTACGATGG  8005927   True  True  SCO7200             2071  \n",
       "1    0.829       GTTAGACTGG  1969219   True  True  SCO1840               88  "
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hrdb_motifs = pd.concat([b_and_p_hrdb_sites, hrdb_downstream_features.loc[:, ['gene', 'dist2downstream']]], axis=1)\n",
    "hrdb_motifs.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "# filter fimo motifs\n",
    "# for each sequence, strand and gene keep motifs closest to downstream and absolute summit\n",
    "#  if those doesn't match, keep one with best p-val\n",
    "selected_fimo_sites = []\n",
    "for _, nn in fimo_motifs.groupby(['sequence_name', 'strand', 'gene']):\n",
    "    if nn.dist2downstream.idxmin() == nn.dist2abs_summit.idxmin():\n",
    "        bmi = nn.dist2downstream.idxmin()\n",
    "    else:\n",
    "        bmi = nn['p-value'].idxmin()\n",
    "        \n",
    "    selected_fimo_sites.append(nn.loc[bmi])\n",
    "selected_fimo_motifs = pd.DataFrame().from_records(selected_fimo_sites)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sequence_name</th>\n",
       "      <th>strand</th>\n",
       "      <th>gene</th>\n",
       "      <th>start_s</th>\n",
       "      <th>stop_s</th>\n",
       "      <th>index_s</th>\n",
       "      <th>start_h</th>\n",
       "      <th>stop_h</th>\n",
       "      <th>index_h</th>\n",
       "      <th>s_h</th>\n",
       "      <th>h_s</th>\n",
       "      <th>cat</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>sigE_101</td>\n",
       "      <td>+</td>\n",
       "      <td>SCO3322</td>\n",
       "      <td>541</td>\n",
       "      <td>566</td>\n",
       "      <td>200</td>\n",
       "      <td>1096</td>\n",
       "      <td>1105</td>\n",
       "      <td>40</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>SHG</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>sigE_101</td>\n",
       "      <td>-</td>\n",
       "      <td>SCO3320</td>\n",
       "      <td>943</td>\n",
       "      <td>968</td>\n",
       "      <td>205</td>\n",
       "      <td>639</td>\n",
       "      <td>648</td>\n",
       "      <td>39</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>SHG</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>sigE_102</td>\n",
       "      <td>+</td>\n",
       "      <td>SCO3203</td>\n",
       "      <td>1198</td>\n",
       "      <td>1224</td>\n",
       "      <td>54322</td>\n",
       "      <td>1203</td>\n",
       "      <td>1212</td>\n",
       "      <td>75</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>O</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>sigE_105</td>\n",
       "      <td>+</td>\n",
       "      <td>SCO3045</td>\n",
       "      <td>287</td>\n",
       "      <td>314</td>\n",
       "      <td>176</td>\n",
       "      <td>414</td>\n",
       "      <td>423</td>\n",
       "      <td>41</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>SHG</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>sigE_105</td>\n",
       "      <td>-</td>\n",
       "      <td>SCO3040</td>\n",
       "      <td>233</td>\n",
       "      <td>258</td>\n",
       "      <td>93</td>\n",
       "      <td>324</td>\n",
       "      <td>333</td>\n",
       "      <td>115</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>HSG</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  sequence_name strand     gene  start_s  stop_s  index_s  start_h  stop_h  \\\n",
       "0      sigE_101      +  SCO3322      541     566      200     1096    1105   \n",
       "1      sigE_101      -  SCO3320      943     968      205      639     648   \n",
       "2      sigE_102      +  SCO3203     1198    1224    54322     1203    1212   \n",
       "3      sigE_105      +  SCO3045      287     314      176      414     423   \n",
       "4      sigE_105      -  SCO3040      233     258       93      324     333   \n",
       "\n",
       "   index_h    s_h    h_s  cat  \n",
       "0       40   True  False  SHG  \n",
       "1       39   True  False  SHG  \n",
       "2       75  False  False    O  \n",
       "3       41   True  False  SHG  \n",
       "4      115  False   True  HSG  "
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sh_merge = pd.merge(\n",
    "    selected_fimo_motifs.loc[:, ['sequence_name', 'strand', 'gene', 'start', 'stop', 'index']], \n",
    "    hrdb_motifs.loc[:, ['sequence_name', 'strand', 'gene', 'start', 'stop', 'index']],\n",
    "    on=['sequence_name', 'strand', 'gene'],\n",
    "    suffixes=['_s', '_h']\n",
    ")\n",
    "sh_merge['s_h'] = False\n",
    "sh_merge.loc[sh_merge.strand == '+', 's_h'] = sh_merge.stop_s < sh_merge.start_h\n",
    "sh_merge.loc[sh_merge.strand == '-', 's_h'] = sh_merge.start_s > sh_merge.stop_h\n",
    "\n",
    "sh_merge['h_s'] = False\n",
    "sh_merge.loc[sh_merge.strand == '+', 'h_s'] = sh_merge.stop_h < sh_merge.start_s\n",
    "sh_merge.loc[sh_merge.strand == '-', 'h_s'] = sh_merge.start_h > sh_merge.stop_s\n",
    "\n",
    "sh_merge['cat'] = 'NAN'\n",
    "sh_merge.loc[sh_merge.s_h, 'cat'] = 'SHG'\n",
    "sh_merge.loc[sh_merge.h_s, 'cat'] = 'HSG'\n",
    "sh_merge.loc[(~sh_merge.s_h) & (~sh_merge.h_s), 'cat'] = 'O'\n",
    "sh_merge.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "# get GC upstream 100\n",
    "ud = 100\n",
    "gc_100 = []\n",
    "for r in selected_fimo_motifs.itertuples():\n",
    "    if r.strand == '+':\n",
    "        s = r.offset + r.start\n",
    "        u = sc_genome.seq[s-ud:s]\n",
    "    elif r.strand == '-':\n",
    "        s = r.offset + r.stop\n",
    "        u = sc_genome.seq[s:s+ud]\n",
    "    else:\n",
    "        raise NotImplementedError()\n",
    "    gc = SeqUtils.GC(u)\n",
    "    gc_100.append(gc)\n",
    "    \n",
    "selected_fimo_motifs['100_upstream_GC'] = gc_100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>motif_id</th>\n",
       "      <th>motif_alt_id</th>\n",
       "      <th>sequence_name</th>\n",
       "      <th>start</th>\n",
       "      <th>stop</th>\n",
       "      <th>strand</th>\n",
       "      <th>score</th>\n",
       "      <th>p-value</th>\n",
       "      <th>q-value</th>\n",
       "      <th>matched_sequence</th>\n",
       "      <th>offset</th>\n",
       "      <th>dist2abs_summit</th>\n",
       "      <th>gene</th>\n",
       "      <th>dist2downstream</th>\n",
       "      <th>100_upstream_GC</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>155</td>\n",
       "      <td>hsm_19</td>\n",
       "      <td>NaN</td>\n",
       "      <td>sigE_0</td>\n",
       "      <td>301</td>\n",
       "      <td>329</td>\n",
       "      <td>-</td>\n",
       "      <td>6.642860</td>\n",
       "      <td>1.890000e-05</td>\n",
       "      <td>0.29900</td>\n",
       "      <td>GGACGAGGTGCTCTGGCCGGCGCGTCTTG</td>\n",
       "      <td>8662666</td>\n",
       "      <td>158</td>\n",
       "      <td>SCO7843</td>\n",
       "      <td>1684.0</td>\n",
       "      <td>67.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>220</td>\n",
       "      <td>hsm_18</td>\n",
       "      <td>NaN</td>\n",
       "      <td>sigE_1</td>\n",
       "      <td>222</td>\n",
       "      <td>249</td>\n",
       "      <td>+</td>\n",
       "      <td>4.750000</td>\n",
       "      <td>3.470000e-05</td>\n",
       "      <td>0.06440</td>\n",
       "      <td>CAACCTCCGGCCATTTCCCACCTTCTGG</td>\n",
       "      <td>8483428</td>\n",
       "      <td>57</td>\n",
       "      <td>SCO7657</td>\n",
       "      <td>103.0</td>\n",
       "      <td>69.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>540</td>\n",
       "      <td>hsm_18</td>\n",
       "      <td>NaN</td>\n",
       "      <td>sigE_1</td>\n",
       "      <td>529</td>\n",
       "      <td>556</td>\n",
       "      <td>+</td>\n",
       "      <td>-0.119048</td>\n",
       "      <td>1.270000e-04</td>\n",
       "      <td>0.10900</td>\n",
       "      <td>GAAGGAGGACCCGGGCGTCTACCTCGAC</td>\n",
       "      <td>8483428</td>\n",
       "      <td>250</td>\n",
       "      <td>SCO7658</td>\n",
       "      <td>1573.0</td>\n",
       "      <td>74.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2604</td>\n",
       "      <td>hsm_16</td>\n",
       "      <td>NaN</td>\n",
       "      <td>sigE_1</td>\n",
       "      <td>529</td>\n",
       "      <td>554</td>\n",
       "      <td>-</td>\n",
       "      <td>-7.559520</td>\n",
       "      <td>9.330000e-04</td>\n",
       "      <td>0.74400</td>\n",
       "      <td>CGAGGTAGACGCCCGGGTCCTCCTTC</td>\n",
       "      <td>8483428</td>\n",
       "      <td>275</td>\n",
       "      <td>SCO7653</td>\n",
       "      <td>3045.0</td>\n",
       "      <td>66.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>24</td>\n",
       "      <td>hsm_18</td>\n",
       "      <td>NaN</td>\n",
       "      <td>sigE_10</td>\n",
       "      <td>154</td>\n",
       "      <td>181</td>\n",
       "      <td>+</td>\n",
       "      <td>19.821400</td>\n",
       "      <td>1.390000e-07</td>\n",
       "      <td>0.00172</td>\n",
       "      <td>GAACGTCATAGCCGTCTGGTGCGTCTAC</td>\n",
       "      <td>7565074</td>\n",
       "      <td>29</td>\n",
       "      <td>SCO6812</td>\n",
       "      <td>9164.0</td>\n",
       "      <td>80.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   index motif_id  motif_alt_id sequence_name  start  stop strand      score  \\\n",
       "0    155   hsm_19           NaN        sigE_0    301   329      -   6.642860   \n",
       "1    220   hsm_18           NaN        sigE_1    222   249      +   4.750000   \n",
       "2    540   hsm_18           NaN        sigE_1    529   556      +  -0.119048   \n",
       "3   2604   hsm_16           NaN        sigE_1    529   554      -  -7.559520   \n",
       "4     24   hsm_18           NaN       sigE_10    154   181      +  19.821400   \n",
       "\n",
       "        p-value  q-value               matched_sequence   offset  \\\n",
       "0  1.890000e-05  0.29900  GGACGAGGTGCTCTGGCCGGCGCGTCTTG  8662666   \n",
       "1  3.470000e-05  0.06440   CAACCTCCGGCCATTTCCCACCTTCTGG  8483428   \n",
       "2  1.270000e-04  0.10900   GAAGGAGGACCCGGGCGTCTACCTCGAC  8483428   \n",
       "3  9.330000e-04  0.74400     CGAGGTAGACGCCCGGGTCCTCCTTC  8483428   \n",
       "4  1.390000e-07  0.00172   GAACGTCATAGCCGTCTGGTGCGTCTAC  7565074   \n",
       "\n",
       "   dist2abs_summit     gene  dist2downstream  100_upstream_GC  \n",
       "0              158  SCO7843           1684.0             67.0  \n",
       "1               57  SCO7657            103.0             69.0  \n",
       "2              250  SCO7658           1573.0             74.0  \n",
       "3              275  SCO7653           3045.0             66.0  \n",
       "4               29  SCO6812           9164.0             80.0  "
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "selected_fimo_motifs.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "def sorter(df):\n",
    "    df['sort'] = df.sequence_name.apply(lambda x: int(x.split('_')[1]))\n",
    "    \n",
    "    return df.sort_values('sort')\\\n",
    "             .drop(columns=['sort'])\\\n",
    "             .rename(columns={'dist2downstream': 'motif2gene_spacer_length'})\n",
    "\n",
    "with pd.ExcelWriter(os.path.join(temp, 'sigE_conserved_motif.xlsx'), engine='xlsxwriter') as writer:\n",
    "    \n",
    "    sheetname = \"sigE_peaks\"\n",
    "    desc = [\n",
    "        \"Columns as defined in sige_macs_peaks_curated.xlsx to allow identification based on 'sequence_name'\",\n",
    "    ]\n",
    "    df_e.to_excel(writer, sheet_name=sheetname, startrow=len(desc)+1, index=False)\n",
    "    for i, r in enumerate(desc):\n",
    "        writer.sheets[sheetname].write(i, 0, r)\n",
    "    \n",
    "    desc = [\n",
    "        \"Motif sites for conserved Tran-like motif constructed from conserved motif sites discovered in sigE peaks sequences using MEME.\",\n",
    "        \"Reported motif sites were found using fimo with p-val threshold of 0.05.\",\n",
    "        \"Filtering:\",\n",
    "        \"  - Duplicated motif sites based on start/stop removed, the one with lower p-value was kept.\",\n",
    "        \"  - In case of multiple motif sites for same downstream gene, the one with lower distance to abs_summit and downstream gene was kept, if ambigous the one with lower p-value was kept.\",\n",
    "        \"Columns:\",\n",
    "        \"  index: motif_site identifier\",\n",
    "        \"  motif_id: motif id (number indicates spacer length)\",\n",
    "        \"  sequence_name: peak sequence id\",\n",
    "        \"  start: motif site start position\",\n",
    "        \"  stop: motif site stop position\",\n",
    "        \"  strand: strand on which motif is found\",\n",
    "        \"  score: fimo match score\",\n",
    "        \"  p-value: motif site P-value (reported by fimo)\",\n",
    "        \"  q-value: motif site q-value (reported by fimo)\",\n",
    "        \"  matched_sequence: discovered motif site sequence 5'->3'\",\n",
    "        \"  offset: sequence start position in genome\",\n",
    "        \"  dist2abs_summit: distance from -10 start to peak summit (absolute value)\",\n",
    "        \"  gene: first gene downstream from motif stop\",\n",
    "        \"  motif2gene_spacer_length: length for a spacer between motif site and downstream gene [nt]\",\n",
    "        \"  100_upstream_GC: GC content in percents for 100nt upstream from -35 start\"\n",
    "    ]\n",
    "    sheetname = 'sigE motif sites'\n",
    "    sorter(selected_fimo_motifs.drop(columns=['motif_alt_id'])).to_excel(writer, sheet_name=sheetname, startrow=len(desc) + 1, index=False)\n",
    "    for i, row in enumerate(desc):\n",
    "        writer.sheets[sheetname].write(i, 0, row)\n",
    "    \n",
    "    desc = [\n",
    "        \"Motif sites for conserved Tran-like motif constructed from conserved motif sites discovered in sigE peaks sequences using MEME.\",\n",
    "        \"Reported motif sites were found using fimo with p-val threshold of 0.05.\",\n",
    "        \"Filtering:\",\n",
    "        \"  - No filtering applied.\",\n",
    "        \"Columns:\",\n",
    "        \"  index: motif_site identifier\",\n",
    "        \"  motif_id: motif id (number indicates spacer length)\",\n",
    "        \"  sequence_name: peak sequence id\",\n",
    "        \"  start: motif site start position\",\n",
    "        \"  stop: motif site stop position\",\n",
    "        \"  strand: strand on which motif is found\",\n",
    "        \"  score: fimo match score\",\n",
    "        \"  p-value: motif site P-value (reported by fimo)\",\n",
    "        \"  q-value: motif site q-value (reported by fimo)\",\n",
    "        \"  matched_sequence: discovered motif site sequence 5'->3'\",\n",
    "        \"  offset: sequence start position in genome\",\n",
    "        \"  dist2abs_summit: distance from -10 start to peak summit (absolute value)\",\n",
    "        \"  gene: first gene downstream from motif stop\",\n",
    "        \"  motif2gene_spacer_length: length for a spacer between motif site and downstream gene [nt]\",\n",
    "    ]\n",
    "    sheetname = 'sigE motif sites unfiltered'\n",
    "    sorter(fimo_motifs_wf.drop(columns=['motif_alt_id'])).to_excel(writer, sheet_name=sheetname, startrow=len(desc) + 1, index=False)\n",
    "    for i, row in enumerate(desc):\n",
    "        writer.sheets[sheetname].write(i, 0, row)\n",
    "    \n",
    "    \n",
    "    \n",
    "    desc = [\n",
    "        \"Best scoring and exact HrdB sites in sigE peaks.\",\n",
    "        \"The sites were found by fimo with HrdB motif, threshold 0.05, Streptomyces coelicolor background. Following motifs sites were selected:\",\n",
    "        \"  1) motif site must contain --TA---T-- pattern\",\n",
    "        \"  2) the best scoring motifs site (by p-value) on each sequence and strand\",\n",
    "        \"Columns:\",\n",
    "        \"  index:       hrdb motif site id\",\n",
    "        \"  motif_id: motif id\",\n",
    "        \"  sequence_name: peak sequence id\",\n",
    "        \"  start: motif site start position\",\n",
    "        \"  stop: motif site stop position\",\n",
    "        \"  strand: strand on which motif is found\",\n",
    "        \"  score: fimo match score\",\n",
    "        \"  p-value: motif site P-value (reported by fimo)\",\n",
    "        \"  q-value: motif site q-value (reported by fimo)\",\n",
    "        \"  matched_sequence: discovered motif site sequence 5'->3'\",\n",
    "        \"  offset: sequence start position in genome\",\n",
    "        \"  exact:            bool wherther motif site was exact match to --TA---T--\",\n",
    "        \"  best:             bool wherther motif site was best match on the sequence (by p-value)\",\n",
    "        \"  gene: first gene downstream from motif stop\",\n",
    "        \"  motif2gene_spacer_length: length for a spacer between motif site and downstream gene [nt]\",\n",
    "    ]\n",
    "    sheetname = 'HrdB motif sites'\n",
    "    sorter(hrdb_motifs).to_excel(writer, sheet_name=sheetname, startrow=len(desc) + 1, index=False)\n",
    "    for i, row in enumerate(desc):\n",
    "        writer.sheets[sheetname].write(i, 0, row)\n",
    "    \n",
    "    \n",
    "    sheetname = 'sigE motif + HrdB motif'\n",
    "    desc = [\n",
    "        \"Intersection of sigE motif sites and HrdB mofif sites.\",\n",
    "        \" '_s' and '_h' suffixes respective for sigE and HrdB motif sites \",\n",
    "        \"Categories:\",\n",
    "        \"  - HSG - order HrdB bs -> sigE bs -> downstream gene\",\n",
    "        \"  - SHG - order sigE bs -> HrdB bs -> downstream gene\",\n",
    "        \"  - O - ambigous - the binding sites ore overlapping each other.\"\n",
    "        \"\"\n",
    "        \"Columns:\",\n",
    "        \"  sequence_name: peak sequence id\",\n",
    "        \"  strand: strand on which motif is found\",\n",
    "        \"  gene: first gene downstream from motif stop\",\n",
    "        \"  start[_s/_h]:  motif site start position\",\n",
    "        \"  stop[_s/_h]:   motif site stop position\",\n",
    "        \"  index[_/_h]:   motif site id\",\n",
    "        \"  cat:  determined category\",\n",
    "    ]\n",
    "    \n",
    "    sorter(sh_merge.drop(columns=['s_h', 'h_s'])).to_excel(writer, sheet_name=sheetname, startrow=len(desc) + 1, index=False)\n",
    "    for i, row in enumerate(desc):\n",
    "        writer.sheets[sheetname].write(i, 0, row)\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.11"
  },
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "state": {},
    "version_major": 2,
    "version_minor": 0
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}