add --ml_method to bypass modelfinder in IQtree (too slow) and update…

… test to use this

funannotate/compare.py

@@ -58,6 +58,7 @@ def __init__(self, prog):
                         help='Pre-computed ProteinOrtho POFF')
     parser.add_argument('--ml_method', default='iqtree',
                         choices=['raxml', 'iqtree'], help='ML method')
+    parser.add_argument('--ml_model', help='IQtree ML model (default: modelfinder)')
     parser.add_argument('-d', '--database',
                         help='Path to funannotate database, $FUNANNOTATE_DB')
     parser.add_argument('--no-progress', dest='progress', action='store_false',
@@ -1232,7 +1233,8 @@ def __init__(self, prog):
             lib.log.info("Inferring phylogeny using RAxML")
             folder = os.path.join(args.out, 'protortho')
             lib.ortho2phylogeny(folder, sco_final, args.num_orthos, busco, args.cpus, args.bootstrap,
-                                phylogeny, outgroup, outgroup_species, outgroup_name, sc_buscos, args.ml_method)
+                                phylogeny, outgroup, outgroup_species, outgroup_name, sc_buscos,
+                                args.ml_method, model=args.ml_model)
             with open(os.path.join(args.out, 'phylogeny.html'), 'w') as output:
                 output.write(lib.HEADER)
                 output.write(lib.PHYLOGENY)

funannotate/funannotate.py

@@ -447,6 +447,7 @@
   --outgroup          Name of species to use for ML outgroup. Default: no outgroup
   --proteinortho      Proteinortho POFF results. in TSV format.
   --ml_method         Maxmimum Liklihood method: Default: iqtree [raxml,iqtree]
+  --ml_model          Substitution model for IQtree. Default: modelfinder
   --no-progress       Do not print progress to stdout for long sub jobs
           """.format(package_name, __version__)
 

funannotate/library.py

@@ -2,6 +2,7 @@
 from __future__ import (absolute_import, division,
                         print_function, unicode_literals)
 from contextlib import contextmanager
+from importlib.util import module_for_loader
 from Bio import BiopythonWarning
 import os
 import uuid
@@ -8284,7 +8285,9 @@ def fasta2dict(Fasta):
     return answer
 
 
-def ortho2phylogeny(folder, df, num, dict, cpus, bootstrap, tmpdir, outgroup, sp_file, name, sc_buscos, ml_method):
+def ortho2phylogeny(folder, df, num, dict, cpus, bootstrap, tmpdir,
+                    outgroup, sp_file, name, sc_buscos, ml_method,
+                    model=False):
     import pylab
     from Bio import Phylo
     from Bio.Phylo.Consensus import get_support
@@ -8353,7 +8356,9 @@ def ortho2phylogeny(folder, df, num, dict, cpus, bootstrap, tmpdir, outgroup, sp
         cmd = ['iqtree', '-s', 'phylogeny.trimal.phylip', '-nt', 'AUTO',
                '-ntmax', str(cpus), '-seed', '12345', '-bb', '1000']
         if outgroup:
-            cmd = cmd + ['-o', name]
+            cmd += ['-o', name]
+        if model:
+            cmd += ['-m', model]
         runSubprocess(cmd, tmpdir, log)
         treefile = os.path.join(tmpdir, 'phylogeny.trimal.phylip.treefile')
         best = Phylo.read(treefile, 'newick')

funannotate/test.py

@@ -290,7 +290,7 @@ def runCompareTest(args):
     # run predict
     runCMD(['funannotate', 'compare',
             '-i', input1, input2, input3,
-            '-o', 'compare', '--cpus', str(args.cpus),
+            '-o', 'compare', '--cpus', str(args.cpus), '--ml_model', 'LG+G4',
             '--outgroup', 'botrytis_cinerea.dikarya'], tmpdir)
     print("#########################################################")
     # check results