added all searchTaxa to phyloprofile

Author: trvinh

fdog/libs/output.py

@@ -90,3 +90,25 @@ def hamstr_2_profile(fa_file):
             for id in list(fa.keys()):
                 tmp = id.split('|')
                 pp.write('%s\tncbi%s\t%s\n' % (tmp[0], tmp[1].split('@')[1], id))
+
+
+def add_all_taxa(pp_file, searchTaxa):
+    """ Add all "missing" search taxa into phyloprofile file """
+    missing_taxa = [] # missing_taxa = [ncbi_id]
+    for taxon in searchTaxa.split(','):
+        flag = general_fn.search_string_in_file(pp_file, taxon)
+        if flag == 0:
+            missing_taxa.append(taxon.split('@')[1])
+    first_gene = ''
+    if os.path.exists(pp_file):
+        with open(pp_file, 'a') as pp:
+            for line in general_fn.read_file(pp_file):
+                if not line.startswith('geneID'):
+                    if not first_gene:
+                        first_gene = line.split('\t')[0]
+                        for i in missing_taxa:
+                            if len(line.split('\t')) == 5:
+                                pp.write(f'{first_gene}\tncbi{i}\tNA\tNA\tNA\n')
+                            else:
+                                pp.write(f'{first_gene}\tncbi{i}\tNA\n')
+                        break

fdog/libs/zzz.py

@@ -181,3 +181,16 @@ def query_yes_no(question, default='yes'):
         else:
             sys.stdout.write('Please respond with "yes" or "no" '
                              '(or "y" or "n").\n')
+
+
+def search_string_in_file(file, string):
+    """ Search for a string in file
+    Return 0 if not found, 1 if found
+    """
+    flag = 0
+    with open(file, 'r') as fp:
+        for l_no, line in enumerate(fp):
+            if string in line:
+                flag = 1
+                break
+    return(flag)

fdog/runMulti.py

@@ -320,7 +320,7 @@ def main():
 
     begin = time.time()
     ##### Check and group parameters
-    print('Preparing & Checking...')
+    print('##### PREPARING & CHECKING #####')
     (inFol, hmmpath, corepath, searchpath, annopath) = prepare_fn.check_input(
                     [inFol, refspec, outpath, hmmpath,
                     corepath, searchpath, annopath, pathFile])
@@ -356,7 +356,7 @@ def main():
 
     ##### DO CORE COMPILATION
     if reuseCore == False:
-        print('Starting compiling core orthologs...')
+        print('##### COMPILING CORE ORTHOLOG GROUPS #####')
         start = time.time()
         coreArgs = [minDist, maxDist, coreSize, coreTaxa, distDeviation,
                     alnStrategy, fasOff]
@@ -382,7 +382,7 @@ def main():
     ##### DO ORTHOLOG SEARCH USING HMM (HAMSTR)
     finalFa = '%s/%s.extended.fa' % (outpath, jobName)
     if not coreOnly:
-        print('Searching orthologs...')
+        print('##### SEARCHING ORTHOLOGS #####')
         start = time.time()
         if not os.path.exists(finalFa) or force == True:
             ### get list of search taxa
@@ -430,6 +430,7 @@ def main():
 
         ##### DO FINAL FAS CALCULATION
         if not fasOff:
+            print('##### CALCULATING FAS SCORES #####')
             try:
                 fasVersion = subprocess.run(['fas.run --version'], shell = True, capture_output = True, check = True)
             except:
@@ -443,6 +444,13 @@ def main():
         else:
             output_fn.hamstr_2_profile(finalFa)
 
+        ##### ADD ALL SEARCH TAXA INTO PhyloProfile OUTPUT
+        pp_file = f'{outpath}/{jobName}.phyloprofile'
+        if not searchTaxa:
+            tmp = general_fn.read_dir(searchpath)
+            searchTaxa = ','.join(tmp)
+        output_fn.add_all_taxa(pp_file, searchTaxa)
+
         end = time.time()
         print('==> fdogs.run finished in ' + '{:5.3f}s'.format(end - begin))
 

fdog/runSingle.py

@@ -23,6 +23,7 @@
 from pkg_resources import get_distribution
 import time
 
+import fdog.libs.zzz as general_fn
 import fdog.libs.preparation as prepare_fn
 import fdog.libs.orthosearch as ortho_fn
 import fdog.libs.corecompile as core_fn
@@ -189,7 +190,7 @@ def main():
 
     begin = time.time()
     ##### Check and group parameters
-    print('Preparing & Checking...')
+    print('##### PREPARING & CHECKING #####')
     if seqFile == 'infile.fa':
         fdogPath = os.path.realpath(__file__).replace('/runSingle.py','')
         seqFile = '%s/data/infile.fa' % fdogPath
@@ -224,7 +225,7 @@ def main():
         seed_id = prepare_fn.get_seed_id_from_fa(core_fa, refspec)
     else:
         seed_id = prepare_fn.identify_seed_id(seqFile, refspec, corepath, debug, silentOff)
-    print('Identified seed ID: %s' % seed_id)
+    print('==> Identified seed ID: %s' % seed_id)
 
     ##### DO CORE COMPILATION
     # start = time.time()
@@ -234,7 +235,7 @@ def main():
                     lowComplexityFilter, evalHmmer/10, coreHitLimit,
                     scoreCutoff, aligner] # rep = True; e-value cutoff is 10x more stringent than from ortho search
     otherCoreArgs = [cpus, debugCore, silentOff, noCleanup, force, append]
-    print('Compiling core set for %s' % seqName)
+    print('##### COMPILING CORE SET FOR %s #####' % seqName)
     core_runtime = core_fn.run_compile_core([seqFile, seqName, refspec, seed_id, reuseCore,
         forceCore, coreArgs, pathArgs, orthoCoreArgs, otherCoreArgs, debug])
     print('==> Core compilation finished in %s' % core_runtime[1])
@@ -243,6 +244,7 @@ def main():
     ##### DO ORTHOLOG SEARCH USING CORE HMM (HAMSTR)
     if not coreOnly:
         start = time.time()
+        print('##### SEARCHING ORTHOLOGS #####')
         # check existing output
         finalOutfile = '%s/%s.extended.fa' % (outpath, seqName)
         finalOutfile = os.path.abspath(finalOutfile)
@@ -278,6 +280,7 @@ def main():
 
         ##### DO FINAL FAS CALCULATION
         if not fasOff:
+            print('##### CALCULATING FAS SCORES #####')
             try:
                 fasVersion = subprocess.run(['fas.run --version'], shell = True, capture_output = True, check = True)
             except:
@@ -290,6 +293,13 @@ def main():
         else:
             output_fn.hamstr_2_profile(finalOutfile)
 
+        ##### ADD ALL SEARCH TAXA INTO PhyloProfile OUTPUT
+        pp_file = f'{outpath}/{seqName}.phyloprofile'
+        if not searchTaxa:
+            tmp = general_fn.read_dir(searchpath)
+            searchTaxa = ','.join(tmp)
+        output_fn.add_all_taxa(pp_file, searchTaxa)
+
         end = time.time()
         print('==> fdog.run finished in ' + '{:5.3f}s'.format(end - begin))
 

setup.py

@@ -26,7 +26,7 @@
 
 setup(
     name="fdog",
-    version="0.1.19",
+    version="0.1.20",
     python_requires='>=3.7.0',
     description="Feature-aware Directed OrtholoG search tool",
     long_description=long_description,