fixed parsing problem of fastANI

moritzbuck · Jun 21, 2021 · d3092d9 · d3092d9
1 parent 044484c
commit d3092d9
Show file tree

Hide file tree

Showing 4 changed files with 10 additions and 5 deletions.
diff --git a/mOTUlizer/__init__.py b/mOTUlizer/__init__.py
@@ -10,7 +10,7 @@
 if os.path.exists(".git"):
     label = subprocess.check_output(["git", "describe", "--tags"]).strip().decode()
 else:
-    label = "0.2.3"
+    label = "0.2.4"
 
 os.chdir(cwd)
 

diff --git a/mOTUlizer/bin/mOTUlize.py b/mOTUlizer/bin/mOTUlize.py
@@ -10,6 +10,7 @@
 import multiprocessing
 from mOTUlizer.classes import *
 from mOTUlizer.utils import *
+from mOTUlizer.config import FASTA_EXTS
 from mOTUlizer.classes.MetaBin import MetaBin
 from mOTUlizer.classes.mOTU import mOTU
 from mOTUlizer import __version__
@@ -28,7 +29,6 @@
 
 """
 
-fasta_exts = [".fna", ".fa", ".fasta", ".fna", ".ffn"]
 
 def motulize(args):
     #parse and check your amino-acid files
@@ -73,8 +73,8 @@ def motulize(args):
             for l in handle:
                 if "query" not in l:
                     ll = l.split("\t")
-                    g1 = ".".join(os.path.basename(ll[0]).split(".")[:-1]) if any([ll[0].endswith(ext) for ext in fasta_exts]) else ll[0]
-                    g2 = ".".join(os.path.basename(ll[1]).split(".")[:-1]) if any([ll[1].endswith(ext) for ext in fasta_exts]) else ll[1]
+                    g1 = ".".join(os.path.basename(ll[0]).split(".")[:-1]) if any([ll[0].endswith(ext) for ext in FASTA_EXTS]) else ll[0]
+                    g2 = ".".join(os.path.basename(ll[1]).split(".")[:-1]) if any([ll[1].endswith(ext) for ext in FASTA_EXTS]) else ll[1]
                     dist = float(ll[2])
                     dist_dict[(g1,g2)] = dist
     else :

diff --git a/mOTUlizer/classes/MetaBin.py b/mOTUlizer/classes/MetaBin.py
@@ -4,6 +4,7 @@
 from subprocess import Popen, PIPE, call
 import os
 import shutil
+from mOTUlizer.config import FASTA_EXTS
 
 class MetaBin:
     def __repr__(self) :
@@ -81,7 +82,10 @@ def get_anis(cls, bins, outfile = None, method = "fastANI", block_size = 500, th
             os.remove(b1_tfile)
             with open(fastani_file) as handle:
                 handle.readline()
-                out_dists = {(l.split()[0], l.strip().split()[1]) : float(l.split()[2]) for l in handle}
+                out_dists = {(os.path.basename(l.split()[0]), os.path.basename(l.strip().split()[1])) : float(l.split()[2]) for l in handle}
+                out_dists = {( ".".join(k[0].split(".")[:-1]) if any([k[0].endswith(ext) for ext in FASTA_EXTS]) else k[0],
+                               ".".join(k[1].split(".")[:-1]) if any([k[1].endswith(ext) for ext in FASTA_EXTS]) else k[1] ): v
+                               for k,v in out_dists.items() }
 #                tfile = lambda k : ".".join(k.split(".")[:-1]) if (k.endswith(".fna") or k.endswith(".fa") or k.endswith(".fasta") or k.endswith(".fna") or k.endswith(".ffn")) else k
 #                out_dists = {(tfile(k[0]),tfile(k[1])) : v for k,v in out_dists.items()}
             if outfile is None:

diff --git a/mOTUlizer/config.py b/mOTUlizer/config.py
@@ -1,2 +1,3 @@
 MIN_PROB = -1000 # log10 of prob of impossible event
 DB_FOLDER = "/home/moritz/dbs/"
+FASTA_EXTS = [".fna", ".fa", ".fasta", ".fna", ".ffn"]