set seqlen limit for muscle5

Author: trvinh

fdog/libs/alignment.py

@@ -72,12 +72,15 @@ def do_align(aligner, fa_file):
     # check muscle version
     if aligner == 'muscle':
         if get_muscle_version(aligner) == 'v3':
-            if fasta_fn.check_long_seq(fa_file) == 1:
+            if fasta_fn.check_long_seq(fa_file, 12000) == 1:
                 aligner = 'mafft-linsi'
             else:
                 aligner = 'muscle_v3'
         else:
-            aligner = 'muscle_v5'
+            if fasta_fn.check_long_seq(fa_file, 15000) == 1:
+                aligner = 'mafft-linsi'
+            else:
+                aligner = 'muscle_v5'
     # create alignment command and run
     align_cline = ''
     if aligner == 'muscle_v3':

fdog/libs/fasta.py

@@ -55,11 +55,12 @@ def append_to_fasta_file(fa_file, new_fa_dict):
                 fa_out.write('>%s\n%s\n' % (id, seq))
 
 
-def check_long_seq(fa_file):
-    """ Check if any sequence longer than 12.000 aa/nt"""
+def check_long_seq(fa_file, max_len):
+    """ Check if any sequence longer than max_len
+    (12.000 aa/nt for muscle v3; 20.000 for muscle v5)"""
     fa_seq = SeqIO.parse(open(fa_file),'fasta')
     for fa in fa_seq:
-        if len(fa.seq) > 12000:
+        if len(fa.seq) > max_len:
             return(1)
     return(0)
 

setup.py

@@ -26,7 +26,7 @@
 
 setup(
     name="fdog",
-    version="0.1.15",
+    version="0.1.16",
     python_requires='>=3.7.0',
     description="Feature-aware Directed OrtholoG search tool",
     long_description=long_description,