Merge pull request #82 from McTavishLab/muscle

Muscle profile alignment

.gitignore

@@ -23,3 +23,7 @@ dist/
 .DS_Store
 ott3.1*
 *.tsv
+physcraper_example_minimal/
+scrape_ot_350_compact/
+physcraper_example_ot_350/
+taxonomy/full_seqs/

docs/examples/data_scrape.py

@@ -5,21 +5,21 @@
 configfi = "docs/examples/example.config"
 study_id = "ot_350"
 tree_id = "Tr53297"
-workdir ="scrape_ot_350_compact"
+workdir ="physcraper_example_ot_350"
 aln_fi = "docs/examples/{}{}.aln".format(study_id, tree_id)
 
 
 # Create an 'scraper' object to get data from NCBI, align it an
-scraper = scraper_from_opentree(study_id = study_id, 
-                                tree_id = tree_id, 
-                                alnfile = aln_fi, 
-                                aln_schema = "nexus", 
-                                configfile = configfi, 
+scraper = scraper_from_opentree(study_id = study_id,
+                                tree_id = tree_id,
+                                alnfile = aln_fi,
+                                aln_schema = "nexus",
+                                configfile = configfi,
                                 workdir = workdir)
 
 sys.stdout.write("{} taxa in alignment and tree\n".format(len(scraper.data.aln)))
 
 
 #scraper.read_blast_wrapper()
 scraper.est_full_tree()
-scraper.data.write_labelled(label='^ot:ottTaxonName')
+scraper.data.write_labelled(label='^ot:ottTaxonName')

docs/examples/minimal.py

@@ -0,0 +1,70 @@
+""" This is a minimal example taking precooked data from the tests
+"""
+import os
+import sys
+import json
+import physcraper
+from physcraper import OtuJsonDict, generate_ATT_from_files, ConfigObj, IdDicts, PhyscraperScrape
+from physcraper import opentree_helpers
+from physcraper.opentree_helpers import scraper_from_opentree
+
+configfi = "tests/data/test.config"
+workdir ="physcraper_example_minimal"
+aln_fi = "tests/data/tiny_test_example/test.fas"
+blast_dir = "tests/data/precooked/fixed/tte_blast_files"
+# mattype="fasta"
+tre_fi= "tests/data/tiny_test_example/test.tre"
+# schema_trf = "newick"
+otu_jsonfi = "{}/otu_dict.json".format(workdir)
+id_to_spn = "tests/data/tiny_test_example/test_nicespl.csv"
+
+if not os.path.exists("{}".format(workdir)):
+    os.makedirs("{}".format(workdir))
+
+conf_base = ConfigObj(configfi)
+
+ids = IdDicts(configfi)
+
+otu_json = OtuJsonDict(id_to_spn, ids)
+with open(otu_jsonfi,"w") as outfile:
+    json.dump(otu_json, outfile)
+
+ottids = [otu_json[ite]['^ot:ottId'] for ite in otu_json]
+
+mrca = opentree_helpers.get_mrca_ott(ottids)
+
+# Create a 'scraper' object to get data from NCBI
+
+data_obj = generate_ATT_from_files(alnfile=aln_fi,
+                             aln_schema="fasta", #mattype
+                             workdir=workdir,
+                             configfile=configfi,
+                             treefile=tre_fi,
+                             tree_schema = "newick", #schema_trf
+                             otu_json=otu_jsonfi,
+                             ingroup_mrca=mrca)
+
+data_obj.tag = "minEx"
+
+scraper = PhyscraperScrape(data_obj, ids)
+
+sys.stdout.write("{} taxa in alignment and tree\n".format(len(scraper.data.aln)))
+
+scraper._blasted = 1 # this tricks PhyScraper into not blasting anything new
+sys.stdout.write("Running read_blast_wrapper()...\n")
+scraper.read_blast_wrapper(blast_dir=blast_dir)
+sys.stdout.write("Running write_aln()...\n")
+aln_path1 = scraper.data.write_aln()
+aln_path_alt = scraper.data.write_aln(filename="already_aligned_seqs.fas")
+unaln_path = scraper.write_new_seqs(filename='unaligned.fas')
+
+sys.stdout.write("Running align_query_seqs()...\n")
+scraper.align_new_seqs()
+scraper.est_full_tree()
+scraper.data.write_labelled(label="^ot:ottTaxonName", norepeats=False)
+scraper.data.write_labelled(label="^ncbi:taxon", filename="ncbi", norepeats=False)
+
+
+# sys.stdout.write("estimating tree...")
+# scraper.est_full_tree()
+# scraper.data.write_labelled(label='^ot:ottTaxonName')

mds/examples.md

@@ -1,4 +1,18 @@
-# Example runs and datasets
+# Example runs
 
-There is a full example python script with comments in `docs/example.py`.
-Some more example files can be found in `docs/example_scripts/`.
+<!-- There is a full example python script with comments in `docs/example.py`.
+Some more example files can be found in `docs/example_scripts/`. -->
+
+## Example ot_350: blasting remotely
+
+Run physcraper interactively on python with the command:
+
+```
+python -i docs/examples/data_scrape.py
+```
+
+Or directly from the command line with:
+
+```
+
+```

mds/testing.md

@@ -9,3 +9,12 @@ To run these tests locally (in your own computer), do
 sh tests/run_test.sh
 sh ws-tests/run_ws-tests.sh`.
 ```
+
+If you want to see each step of the physcraper algorithm in action, you can also
+run part of the tests in a python session.
+
+For example, you can do:
+
+```
+python test/test_multi.py
+```

physcraper/aligntreetax.py

@@ -573,26 +573,31 @@ def write_random_resolve_tre(self, treefilename='random_resolve.tre'):
         fi.close()
         return treepath
 
-    def write_aln(self, alnname=None, alnschema="fasta"):
-        if alnname == None:
-            alnname = "physcraper_{}.fas".format(self.tag)
-        alnpath = "{}/{}".format(self.workdir, alnname)
+    def write_aln(self, filename=None, alnschema="fasta"):
+        if filename == None:
+            filename = "physcraper_{}.fas".format(self.tag)
+        alnpath = "{}/{}".format(self.workdir, filename)
         self.aln.write(path=alnpath,
                        schema=alnschema)
         return os.path.abspath(alnpath)
 
-    def write_files(self, treepath=None, treeschema="newick", alnpath=None, alnschema="fasta"):
+    def write_files(self, treefilename=None, treeschema="newick", alnfilename=None, alnschema="fasta"):
         """Outputs both the streaming files, labeled with OTU ids.
         Can be mapped to original labels using otu_dict.json or otu_seq_info.csv"""
         #debug("write_files")
-        if alnpath == None:
-            alnpath = "physcraper_{}.fas".format(self.tag)
-        if treepath == None:
+        if treefilename == None:
             treepath = "physcraper_{}.tre".format(self.tag)
-        self.tre.write(path="{}/{}".format(self.workdir, treepath),
+        else:
+            treepath = "{}/{}".format(self.workdir, treefilename)
+        if alnfilename == None:
+            alnpath = "physcraper_{}.fas".format(self.tag)
+        else:
+            alnpath = "{}/{}".format(self.workdir, alnfilename)
+        self.tre.write(path=treepath,
                        schema=treeschema, unquoted_underscores=True)
-        self.aln.write(path="{}/{}".format(self.workdir, alnpath),
+        self.aln.write(path=alnpath,
                        schema=alnschema)
+        return(treepath, alnpath)
 
 
     def write_labelled(self, label, filename = "labelled", direc='workdir', norepeats=True, add_gb_id=False):