various cleanup

cms/combine/cms_data.c

@@ -1,5 +1,5 @@
 // functions for handling cms component(+composite) score datastructures
-// last updated: 1.10.17 	vitti@broadinstitute.org
+// last updated: 2.10.2017 	vitti@broadinstitute.org
 
 #include <stdio.h>
 #include <string.h>
@@ -301,7 +301,7 @@ void get_ihs_data(ihs_data* data, char filename[]) {
 			else if (itoken == 5) {
 				data->ihs_unnormed[isnp] = atof(token);
 			}			
-			else if (itoken == 10) {
+			else if (itoken == 6) {
 				data->ihs_normed[isnp] = atof(token);
 			}
 			//else if (itoken == 7) {

cms/combine/input_func.py

@@ -0,0 +1,56 @@
+## functions for composite
+## last updated: 	02.10.2017	vitti@broadinstitute.org
+
+import sys
+
+def get_likesfiles_frommaster(masterfile, modelselpop, vs="neut"):
+	"""assumes only one demographic model in master likes file, but there may be multiple pops or neut/linked comps"""
+	likesfiles = []
+	openfile = open(masterfile, 'r')
+	for line in openfile:
+		filename = line.strip('\n')
+		likesfiles.append(filename)
+	openfile.close()
+
+	print('master likesfile has a total of ' + str(len(likesfiles)) + " files..")
+	#print(likesfiles)
+	selpop_likesfiles = [item for item in likesfiles if "sel" + str(modelselpop) in item]
+	#print(selpop_likesfiles)
+	ihsfiles = [item for item in selpop_likesfiles if "/ihs/" in item]
+	nslfiles = [item for item in selpop_likesfiles if "/nsl/" in item]
+	delihhfiles = [item for item in selpop_likesfiles if "/delihh/" in item]
+	xpehhfiles = [item for item in selpop_likesfiles if "/xpehh/" in item]
+	fstfiles = [item for item in selpop_likesfiles if "/fst/" in item]
+	deldaffiles = [item for item in selpop_likesfiles if "/deldaf/" in item]
+
+	ihs_selfiles = [item for item in ihsfiles if "causal" in item]
+	nsl_selfiles = [item for item in nslfiles if "causal" in item]
+	delihh_selfiles = [item for item in delihhfiles if "causal" in item]
+	xpehh_selfiles = [item for item in xpehhfiles if "causal" in item]
+	fst_selfiles = [item for item in fstfiles if "causal" in item]
+	deldaf_selfiles = [item for item in deldaffiles if "causal" in item]
+
+	ihs_neutfiles = [item for item in ihsfiles if vs in item]
+	nsl_neutfiles = [item for item in nslfiles if vs in item]
+	delihh_neutfiles = [item for item in delihhfiles if vs in item]
+	xpehh_neutfiles = [item for item in xpehhfiles if vs in item]
+	fst_neutfiles = [item for item in fstfiles if vs in item]
+	deldaf_neutfiles = [item for item in deldaffiles if vs in item]
+
+	for filelist in [ihs_selfiles, ihs_neutfiles, nsl_selfiles, nsl_neutfiles, delihh_selfiles, delihh_neutfiles, xpehh_selfiles, xpehh_neutfiles, fst_selfiles, fst_neutfiles, deldaf_selfiles, deldaf_neutfiles]:
+		#print(filelist)
+		if len(filelist) >1:
+			print('conflicting distributions found...')
+			sys.exit()
+		elif len(filelist) == 0:
+			print('missing distributions for population ' + str(modelselpop))
+			sys.exit()
+
+	else:
+		ihs_hit_filename, ihs_miss_filename = ihs_selfiles[0], ihs_neutfiles[0]
+		nsl_hit_filename, nsl_miss_filename = nsl_selfiles[0], nsl_neutfiles[0]
+		delihh_hit_filename, delihh_miss_filename = delihh_selfiles[0], delihh_neutfiles[0]
+		xpehh_hit_filename, xpehh_miss_filename = xpehh_selfiles[0], xpehh_neutfiles[0]	
+		fst_hit_filename, fst_miss_filename = fst_selfiles[0], fst_neutfiles[0]	
+		deldaf_hit_filename, deldaf_miss_filename = deldaf_selfiles[0], deldaf_neutfiles[0]
+	return ihs_hit_filename, ihs_miss_filename, nsl_hit_filename, nsl_miss_filename, delihh_hit_filename, delihh_miss_filename,  xpehh_hit_filename, xpehh_miss_filename, fst_hit_filename, fst_miss_filename, deldaf_hit_filename, deldaf_miss_filename

cms/combine/likes_func.py

@@ -1,5 +1,5 @@
 ## functions for composite
-## 11.24.16	vitti@broadinstitute.org
+## last updated: 	02.10.2017	vitti@broadinstitute.org
 
 import sys
 
@@ -54,5 +54,27 @@ def get_likesfiles_frommaster(masterfile, modelselpop, vs="neut"):
 		fst_hit_filename, fst_miss_filename = fst_selfiles[0], fst_neutfiles[0]	
 		deldaf_hit_filename, deldaf_miss_filename = deldaf_selfiles[0], deldaf_neutfiles[0]
 	return ihs_hit_filename, ihs_miss_filename, nsl_hit_filename, nsl_miss_filename, delihh_hit_filename, delihh_miss_filename,  xpehh_hit_filename, xpehh_miss_filename, fst_hit_filename, fst_miss_filename, deldaf_hit_filename, deldaf_miss_filename
-
-
+def write_pair_sourcefile(writefilename, ihsfilename, delihhfilename, nslfilename, xpehhfilename, freqsfilename):
+	if not os.path.isfile(writefilename):
+		openfile = open(writefilename, 'w')
+		openfile.write(ihsfilename+ "\n")
+		openfile.write(delihhfilename+ "\n")
+		openfile.write(nslfilename+ "\n")
+		openfile.write(xpehhfilename+ "\n")
+		openfile.write(freqsfilename+ "\n")
+		openfile.close()
+	return writefilename
+def write_run_paramfile(writefilename, ihs_master_likesfile, nsl_master_likesfile, delihh_master_likesfile, xpehh_master_likesfile,
+	fst_master_likesfile, deldaf_master_likesfile, cutoffline, includeline):
+	if not os.path.isfile(writefilename):
+		openfile = open(writefilename, 'w')
+		openfile.write(ihs_master_likesfile + "\n")
+		openfile.write(nsl_master_likesfile + "\n") #CHANGE ORDER
+		openfile.write(delihh_master_likesfile + "\n")
+		openfile.write(xpehh_master_likesfile + "\n")
+		openfile.write(fst_master_likesfile + "\n")
+		openfile.write(deldaf_master_likesfile + "\n")	
+		openfile.write(cutoffline + "\n")
+		openfile.write(includeline + "\n")		
+		openfile.close()
+	return writefilename

cms/composite.py

@@ -1,12 +1,11 @@
 #!/usr/bin/env python
 ## top-level script for combining scores into composite statistics as part of CMS 2.0.
-## last updated: 12.16.16 	vitti@broadinstitute.org
+## last updated: 02.10.2017 	vitti@broadinstitute.org
 
 import matplotlib
 matplotlib.use('agg')
 import matplotlib.pyplot as plt
-from combine.recalc_func import interpolate_haps, windows, interpolate_from_windows
-from combine.likes_func import get_likesfiles_frommaster
+from combine.input_func import get_likesfiles_frommaster, write_pair_sourcefile, write_run_paramfile
 from combine.viz_func import hapSort_coreallele, hapSort, hapViz, readAnnotations, find_snp_index, pullRegion, load_from_hap
 from dists.scores_func import calc_fst_deldaf, calc_delihh
 import subprocess
@@ -179,14 +178,6 @@ def execute_win_haps(args):
 	writefilename = args.writefilename
 	windows(windowsize, jumplen, infilename, writefilename)
 	return
-def execute_interpolate_hapscores(args):
-	''' strike '''
-	inputTpedFilename = args.intpedfilename
-	inputIhsFilename = args.inihsfilename
-	inputWinihsFilename = args.inwinihsfilename
-	outputFilename = args.outfilename
-	interpolate_from_windows(inputTpedFilename, inputIhsFilename, inputWinihsFilename, outputFilename)
-	return
 def execute_delihh_from_ihs(args):
 	calc_delihh(args.readfile, args.writefile)
 	return
@@ -203,31 +194,10 @@ def execute_xp_from_ihh(args):
 	else:
 		subprocess.check_call( cmdstring.split() )	
 	return
-def execute_poppair(args):
-	if args.cmsdir is not None:
-		cmd = args.cmsdir
-	else:
-		cmd = ""
-	cmd += "combine/combine_scores_poppair"
-	if args.xp_reverse_pops:
-		xp_reversed = 0
-	else:
-		xp_reversed = 1
-	if args.fst_deldaf_reverse_pops:
-		deldaf_reversed = 0
-	else:
-		deldaf_reversed = 1
-	argstring = args.in_ihs_file + " "  + args.in_nsl_file + " " + args.in_delihh_file + " " + args.in_xp_file + " " + str(xp_reversed) + " " + args.in_fst_deldaf_file + " " + str(deldaf_reversed) + " " + args.outfile 
-	cmdstring = cmd + " " + argstring
-	if args.printOnly:
-			print(cmdstring)
-	else:
-		subprocess.check_call( cmdstring.split() )	
-	return
 def execute_outgroups(args):
-
+	''' nix this? execute_composite in power.py '''
+	''' composites scores from input components and likelihood tables '''
 	ihs_hit_hi_filename, ihs_miss_hi_filename, nsl_hit_hi_filename, nsl_miss_hi_filename, delihh_hit_hi_filename, delihh_miss_hi_filename, xpehh_hit_hi_filename, xpehh_miss_hi_filename, fst_hit_hi_filename, fst_miss_hi_filename, deldaf_hit_hi_filename, deldaf_miss_hi_filename = get_likesfiles_frommaster(args.likesfile, args.selpop_likes)
-
 	usefreqs = []
 	if args.likesfile_low is not None:
 		ihs_hit_low_filename, ihs_miss_low_filename, nsl_hit_low_filename, nsl_miss_low_filename, delihh_hit_low_filename, delihh_miss_low_filename, xpehh_hit_low_filename, xpehh_miss_low_filename, fst_hit_low_filename, fst_miss_low_filename, deldaf_hit_low_filename, deldaf_miss_low_filename = get_likesfiles_frommaster(args.likesfile_low, args.selpop_likes)
@@ -242,23 +212,22 @@ def execute_outgroups(args):
 		cmd = args.cmsdir
 	else:
 		cmd = ""
+	cmd += "combine/combine_scores"
+	argstring = args.outfile 
 
-
-	if not args.region: 	#GENOME-WIDE
-		cmd += "combine/combine_scores_multiplepops_filtermaf"
-		argstring = args.outfile 
-	else:	#WITHIN REGION
-		cmd += "combine/combine_scores_multiplepops_region"
-		argstring = args.outfile + " " + str(args.startBp) + " " + str(args.endBp) + " "  
-
-	for score in ['ihs', 'nsl', 'delihh', 'xpehh', 'fst', 'deldaf']:
-		for dist_type in ['hit', 'miss']:
-			for freq in usefreqs: #['low', 'mid', 'hi']:
-				argument = eval(score + "_" + dist_type + "_" + freq + "_filename")
-				argstring += " " + argument 
+	#for score in ['ihs', 'nsl', 'delihh', 'xpehh', 'fst', 'deldaf']:
+	#	for dist_type in ['hit', 'miss']:
+	#		for freq in usefreqs: #['low', 'mid', 'hi']:
+	#			argument = eval(score + "_" + dist_type + "_" + freq + "_filename")
+	#			argstring += " " + argument 
+
+	argstring += " " + cms_run_paramfilename
 
+
+	#write_pair_sourcefile(writefilename, ihsfilename, delihhfilename, nslfilename, xpehhfilename, freqsfilename):
 	for pairfile in args.infiles.split(','):
 		argstring += " " + pairfile
+
 	cmdstring = cmd + " " + argstring
 	if args.printOnly:
 			print(cmdstring)