def parseOnlyEffScores(inFname):
""" return a dict seq -> scoreType -> score
"""
print "reading %s" % inFname
scores = {}
#freqs = {}
for row in annotateOffs.iterTsvRows(inFname):
seq = row.seq
scores[seq] = {}
dataDict = row._asdict()
for st in annotateOffs.scoreTypes:
scores[seq][st] = float(dataDict[st])
guideSeq = row.seq[:20]
scores[seq]["finalGc6"] = int(annotateOffs.countFinalGc(guideSeq, 6)>=4)
scores[seq]["finalGg"] = int(guideSeq[-2:]=="GG")
scores[seq]["modFreq"] = float(row.modFreq)
#freq = float(row.modFreq)
#freqs[seq] = freq
assert(len(scores)!=0)
return scores
#import rpy
#import rpy2.robjects as ro
fnames = ["effData/concordet2.scores.tab", "effData/alenaAll.scores.tab", \
"effData/teboulVivo_mm9.scores.tab", "effData/schoenig.scores.tab", "effData/eschstruth.scores.tab"]
#corrOverlap = ro.r['cocor.dep.groups.overlap']
ofh = open("out/signFig5.tab", "w")
for fname in fnames:
x1 = []
x2 = []
y = []
#for scoreType in ["fusi", "crisprScan"]:
for row in annotateOffs.iterTsvRows(fname):
x1.append(float(row.fusi))
x2.append(float(row.crisprScan))
y.append(float(row.modFreq))
r1 = stats.pearsonr(x1, y)[0]
r2 = stats.pearsonr(x2, y)[0]
r12 = stats.pearsonr(x1, x2)[0]
#res = corrOverlap(r.jk=r1, r.jh=r2, r.kh=r12, n=len(x1), alternative="less", alpha=0.05, conf.level=0.95, null.value=0)
if r1 < r2:
way = "less"
else:
way = "more"
row = [fname, r1, r2, r12, len(x1), way]
row = [str(x) for x in row]
ofh.write("\t".join(row) + "\n")
#import rpy
#import rpy2.robjects as ro
fnames = ["effData/concordet2.scores.tab", "effData/alenaAll.scores.tab", \
"effData/teboulVivo_mm9.scores.tab", "effData/schoenig.scores.tab", "effData/eschstruth.scores.tab"]
#corrOverlap = ro.r['cocor.dep.groups.overlap']
ofh = open("out/signFig5.tab", "w")
for fname in fnames:
x1 = []
x2 = []
y = []
#for scoreType in ["fusi", "crisprScan"]:
for row in annotateOffs.iterTsvRows(fname):
x1.append(float(row.fusi))
x2.append(float(row.crisprScan))
y.append(float(row.modFreq))
r1 = stats.pearsonr(x1, y)[0]
r2 = stats.pearsonr(x2, y)[0]
r12 = stats.pearsonr(x1, x2)[0]
#res = corrOverlap(r.jk=r1, r.jh=r2, r.kh=r12, n=len(x1), alternative="less", alpha=0.05, conf.level=0.95, null.value=0)
if r1 < r2:
way = "less"
else:
way = "more"
row = [fname, r1, r2, r12, len(x1), way]
row = [str(x) for x in row]
ofh.write("\t".join(row)+"\n")
