Repository navigation
UPP2
Angus Ball edited this page Oct 8, 2026
·
6 revisions
Why does UPP2 get it's own page? idk Im afraid. \
We need
Python: Version >= 3.6. SEPP: Version > 3.0. PASTA: Version > 1.0.
Sepp comes with UPP2, so all we need to explicitly download is PASTA and some random packages
pip install numpy==1.21.6
pip install dendropy==4.5.2
First try:
conda install bioconda::pasta
conda install -c bioconda pplacer
Upp2 we need to download from first principles, SINCE we are using a fork from the main package
mkdir Upp2
cd Upp2
git clone https://github.com/gillichu/sepp.git
Might have to move into the git folder
python setup.py develop --user
From the UPP2 wiki: \
[commandline] sequence_file= alignment= tree= alignmentSize= molecule=<amino/dna/rna> cpu=16 tempdir=./tmp/ outdir=./output/
[upp2] decomp_only=True bitscore_adjust=True hier_upp=True early_stop=<True/False>
The bare code would be:
python run_upp.py -c <Config file>
Where the config file looks like
[commandline]
sequence_file=<query sequence file>
alignment=<backbone alignment>
tree=<backbone tree>
alignmentSize=<decomposition size>
molecule=<amino/dna/rna>
cpu=16
tempdir=./tmp/
outdir=./output/
[upp2]
decomp_only=True
bitscore_adjust=True
hier_upp=True
early_stop=<True/False>
Early stop is really important as it is whats actually getting us the runtime bonus
alignmentSize=, is set to 10 or 2, 2 increasing runtime, 10 saw no increase in alignment error, using 10 \
Therefore the actual code is... \
python run_upp.py -c upp2_greengenes_config.file
where the config file is
[commandline]
sequence_file=Oopsall_asvs_norc.fasta
alignment=greengenes2_MSA.fasta
tree=greengenes2_tree.nwk
alignmentSize=10
molecule=dna
cpu=30
tempdir=./tmp/
outdir=./output_greengenes/
checkpoint=./output_greengenes/test_500.chk
[upp2]
decomp_only=True
bitscore_adjust=True
hier_upp=True
early_stop=True
[exhaustive]
strategy = centroid
minsubsetsize = 2
placementminsubsetsizefacotr = 4
placer = pplacer
weight_placement_by_alignment = True
[pplacer]
path=/home/coder/project/R/nonR_libraries/miniconda3/envs/py310/bin/pplacer
[hmmalign]
path=/home/coder/project/R/nonR_libraries/miniconda3/envs/py310/bin/hmmalign
[hmmsearch]
path=/home/coder/project/R/nonR_libraries/miniconda3/envs/py310/bin/hmmsearch
piped=False
elim=10000
filters=True
[hmmbuild]
path=/home/coder/project/R/nonR_libraries/miniconda3/envs/py310/bin/hmmbuild
[jsonmerger]
path=./tools/merge/seppJsonMerger.jar
All the extra bs hmmbuild stuff is just links to where those commands can be found \
python run_upp.py --help
[07:35:34] config.py (line 337): WARNING: Main configuration file was not found at: /home/coder/project/R/nonR_libraries/UPP2/sepp/.sepp/upp.config
Proceeding without the main configuration...
usage: run_upp.py [--help] [-v] [-F N] [-D DISTANCE] [--diameter DIAMETER] [-p DIR] [-rt] [-o OUTPUT] [-d OUTPUT_DIR] [--molecule MOLECULE]
[--ignore-overlap] [-x N] [-cp CHCK_FILE] [-cpi N] [-seed N] [-A N] [-R "Nmin Nmax"] [-M N] [-T N] [-B N] [-S DECOMP]
[-s SEQ] [-c CONFIG] [-t TREE] [-a ALIGN] [-b SEQ] [-l N] [-j DCOMP] [-h HIER] [-z BSADJUST] [-e EARLYSTOP]
[-m ENSEMBLEPATH] [-P N] [-r RAXML] [-f FRAG]
This script runs the UPP algorithm on set of sequences. A backbone alignment and tree can be given as input. If none is provided, a backbone
will be automatically generated.
options:
--help show this help message and exit
-v, --version show program's version number and exit
DECOMPOSITION OPTIONS:
These options determine the alignment decomposition size, backbone size, and how to decompose the backbone set.
-F N, --fragmentChunkSize N
maximum fragment chunk size of N. Helps controlling memory. [default: 999999999]
-D DISTANCE, --distance DISTANCE
minimum p-distance before stopping the decomposition[default: 1]
--diameter DIAMETER maximum tree diameter before stopping the decomposition[default: None]
-A N, --alignmentSize N
max alignment subset size of N [default: 10]
-R "Nmin Nmax", --full_length_range "Nmin Nmax"
Only consider sequences with lengths within Nmin and Nmax
-M N, --median_full_length N
Consider all fragments that are 25% longer or shorter than N to be excluded from the backbone. If value is -1, then
UPP will use the median of the sequences as the median full length [default: None]
-T N, --backbone_threshold N
Only consider sequences with lengths within (1-N)*M and (1+N)*M as full-length, where M is the median length of the
full-length sequence given by the -M option and N is the percentage given by the -T option.[default: 0.25]
-B N, --backboneSize N
(Optional) size of backbone set. If no backbone tree and alignment is given, the sequence file will be randomly
split into a backbone set (size set to N) and query set (remaining sequences), [default: min(1000,input size)]
-S DECOMP, --decomp_strategy DECOMP
decomposition strategy [default: ensemble of HMMs (hierarchical)]
OUTPUT OPTIONS:
These options control output.
-p DIR, --tempdir DIR
Tempfile files will be written to DIR. Full-path required. [default: /tmp/sepp]
-rt, --remtemp Remove tempfile directory. [default: disabled]
-o OUTPUT, --output OUTPUT
output files with prefix OUTPUT. [default: output]
-d OUTPUT_DIR, --outdir OUTPUT_DIR
output to OUTPUT_DIR directory. full-path required. [default: .]
INPUT OPTIONS:
These options control input. To run SEPP the following is required. A backbone tree (in newick format), a RAxML_info file (this is the
file generated by RAxML during estimation of the backbone tree. Pplacer uses this info file to set model parameters), a backbone
alignment file (in fasta format), and a fasta file including fragments. The input sequences are assumed to be DNA unless specified
otherwise.
--molecule MOLECULE Molecule type of sequences. Can be amino, dna, or rna [default: dna]
--ignore-overlap When a query sequence has the same name as a backbone sequence, ignore the query sequences and keep the backbone
sequence[default: False]
-s SEQ, --sequence_file SEQ
Unaligned sequence file. If no backbone tree and alignment is given, the sequence file will be randomly split into a
backbone set (size set to B) and query set (remaining sequences), [default: None]
-c CONFIG, --config CONFIG
A config file, including options used to run UPP. Options provided as command line arguments overwrite config file
values for those options. [default: None]
-t TREE, --tree TREE Input tree file (newick format) [default: None]
-a ALIGN, --alignment ALIGN
Aligned fasta file [default: None]
-b SEQ, --backtranslation SEQ
Fasta file containing unaligned DNA sequences corresponding every reference and query sequence [default: None]
OTHER OPTIONS:
These options control how SEPP is run
-x N, --cpu N Use N cpus [default: number of cpus available on the machine]
-cp CHCK_FILE, --checkpoint CHCK_FILE
checkpoint file [default: no checkpointing]
-cpi N, --interval N Interval (in seconds) between checkpoint writes. Has effect only with -cp provided. [default: 3600]
-seed N, --randomseed N
random seed number. [default: 297834]
UPP OPTIONS:
These options set settings specific to UPP
-l N, --longbranchfilter N
Branches longer than N times the median branch length are filtered from backbone and added to fragments. [default:
None (no filtering)]
UPP2 OPTIONS:
These options set settings specific to UPP2
-j DCOMP, --decomp_only DCOMP
Only run the decomposition step of UPP. [default: False]
-h HIER, --hier_upp HIER
Run the hierarchical search through the UPP HMMs. [default: False]
-z BSADJUST, --bitscore_adjust BSADJUST
Run with adjusted bitscore weighting. [default: False]
-e EARLYSTOP, --early_stop EARLYSTOP
Run with early stop in hierarchical search. [default: False]
-m ENSEMBLEPATH, --ensemble_path ENSEMBLEPATH
Path to ensemble data from a previous run of UPP2[default: False]
SEPP OPTIONS:
These options set settings specific to SEPP and are not used for UPP.
-P N, --placementSize N
max placement subset size of N [default: 10% of the total number of taxa]
-r RAXML, --raxml RAXML
RAxML_info file including model parameters, generated by RAxML.[default: None]
-f FRAG, --fragment FRAG
fragment file [default: None]