def getFastaCount( fasta, intervals ):
f = open( fasta )
counts = []
for i in range( len(intervals) -1 ):
counts.append(0)
for r in f:
if r.strip()[0] != '>':
continue
v = int(r.split('=')[-1])
idx = getIntervalIdx( intervals, v)
if idx != None:
counts[ idx ] += 1
return counts
def main( args ):
records, motifs, details_poses = getRecords( args.input )
args.intervals = list(set(args.intervals))
args.intervals.sort()
if len( args.intervals ) < 2:
print "Intervals not defined."
return
result = []
distBtMotifs = []
leftPoses = []
rightPoses = []
singlePoses = []
for i in range( len(motifs) ):
temp = []
distTemp = []
for j in range( len(args.intervals) - 1 ):
temp.append({})
distTemp.append([])
result.append(temp)
distBtMotifs.append( distTemp )
leftPoses.append([])
rightPoses.append([])
singlePoses.append([])
if len(records) < 1 : sys.exit()
motifIndex = 0
currMotif = motifs[motifIndex]
currSeqName = records[0][0]
currSeqMotifCount = 0
intervalIndex = getIntervalIdx( args.intervals, records[0][2])
print args.intervals
for r in records:
detail_r = details_poses[r[-1]][r[0]]
if len(detail_r) > 1:
if intervalIndex != None:
tempDist = detail_r[-1][0] - detail_r[0][0]
distBtMotifs[ motifIndex ] [intervalIndex].append( tempDist )
leftPoses[ motifIndex ].append( getMotifCenterPos( detail_r[0][0], detail_r[0][1] ) )
rightPoses[ motifIndex ].append( getMotifCenterPos( detail_r[-1][0], detail_r[-1][1] ) )
else:
singlePoses[ motifIndex ].append( getMotifCenterPos( detail_r[0][0], detail_r[0][1] ) )
if r[-1] != currMotif:
motifIndex += 1
assert r[-1] == motifs[motifIndex]
currMotif = motifs[motifIndex]
else:
currSeqMotifCount = len( details_poses[r[-1]][r[0]] )
#print r[2],' ', intervalIndex
if intervalIndex != None and currSeqMotifCount in result[motifIndex][intervalIndex]:
result[motifIndex][intervalIndex][ currSeqMotifCount ] += 1
elif intervalIndex != None:
result[motifIndex][intervalIndex][ currSeqMotifCount ] = 1
currSeqMotifCount = 1
intervalIndex = getIntervalIdx( args.intervals, r[2] )
currSeqName = r[0]
fastaCount = getFastaCount( args.fasta, args.intervals )
translate = loadTranslate( args.translate )
counts = printResult( result, motifs, args.intervals, args.outPrefix, fastaCount, translate)
printDistBox( distBtMotifs, args.intervals, motifs, translate, args.plotsdir)
printLocHist( leftPoses, rightPoses,singlePoses, motifs, translate, args.plotsdir )
generateHeatMatrix( motifs, details_poses, records, 201, args.intervals, args.plotsdir)
plotPieChart(counts, fastaCount, args.plotsdir)
def generateHeatMatrix( motifs, details, records, seq_length, intervals , savefig=None):
import numpy as np
import pylab as pl
from plotDistCat import saveOrPrint, getIntervalIdx
data = {}
motif2value = {}
for i,m in enumerate(motifs):
print m
motif2value[ m ] = i
print motifs
for r in records:
if r[-1] not in motif2value:
continue
if r[0] not in data:
data[r[0]] = [r[2],]
for i in range(len(motifs)):
data[r[0]].append([])
for p in details[r[-1]][r[0]]:
data[r[0]][motif2value[r[-1]]+1].append(( p[0],p[1]))
matrix_all = []
matrix_exist = []
ticks = [0, seq_length/2, seq_length]
labels = [-(seq_length/2),0,seq_length/2]
count_diff = 0
count_all = 0
count_all_by_dist = []
count_exist_by_dist = []
#the following are used to store the indexes of the rows in
# the complete set
matrix_by_dist = []
matrix_exist_by_dist = []
for i in range(len(intervals) - 1):
matrix_by_dist.append([])
matrix_exist_by_dist.append([])
count_all_by_dist.append(0)
count_exist_by_dist.append(0)
NUMPLOTS = len(intervals)
for seq in data:
exist = True
temp = np.zeros(seq_length) - 1
for i,p in enumerate(data[seq][1:]):
for t in p:
temp[(t[0]+t[1])/2] = i
if len(p) == 0:
exist = False
temp = list(temp)
if sum(temp) == -1*seq_length:
continue
interIdx = getIntervalIdx(intervals, data[seq][0])
matrix_all.append(temp)
if interIdx != None:
matrix_by_dist[interIdx].append(len(matrix_all)-1)
count_all_by_dist[interIdx] += 1
count_all += 1
if exist:
matrix_exist.append(temp)
if interIdx != None:
matrix_exist_by_dist[interIdx].append(len(matrix_exist)-1)
count_exist_by_dist[interIdx] += 1
else:
count_diff += 1
print count_diff, ' ', count_all, ' ', len(matrix_all),' ', len(matrix_exist)
print count_all_by_dist
print count_exist_by_dist
pl.close("all")
fig_all = pl.figure(1)
fig_exist = pl.figure(2)
sub_count = 1
matrix_all_idx = range(len(matrix_all))
matrix_exist_idx = range(len(matrix_exist))
matrix_all_idx.sort(key=lambda k:(np.average(matrix_all[k]),))
matrix_exist_idx.sort(key=lambda k:(np.average(matrix_exist[k]),))
for i in range(len(intervals) - 1):
matrix_by_dist[i].sort(key=lambda k:(np.average(matrix_all[k]),))
matrix_exist_by_dist[i].sort(key=lambda k:(np.average(matrix_exist[k]),))
subTitles = []
for i in range(len(intervals) - 1):
t = '[%d, %d'%(intervals[i], intervals[i+1])
if i != len(intervals) - 2:
t += ')'
else:
t += ']'
subTitles.append(t)
#matrix_all.sort(key=lambda k:(np.average(k),)) # np.average(k[0:60]), np.average(k[60:140]),np.average(k[140:])))
#matrix_exist.sort(key=lambda k:(np.average(k),)) # np.average(k[0:60]), np.average(k[60:140]),np.average(k[140:])))
matrix_all = np.array(matrix_all)
matrix_exist = np.array(matrix_exist)
#matrix_all[matrix_all<0] = None
#matrix_exist[matrix_exist<0] = None
matrix_all += 1
matrix_exist += 1
ncolumns = 2
def setTickAndLabel(ax, ticks_a, labels_a):
ax.set_xticks(ticks_a)
ax.set_xticklabels(labels_a)
if len(intervals) > 2:
for i in range(len(intervals) - 1):
fas = fig_all.add_subplot((NUMPLOTS+1)/2, 2, sub_count)
fes = fig_exist.add_subplot((NUMPLOTS+1)/2, 2, sub_count)
#fas.imshow(matrix_all[matrix_by_dist[i]], aspect='auto')
fas.plot(matrix_all[matrix_by_dist[i]].sum(axis=0)/matrix_all[matrix_by_dist[i]].shape[0])
fas.set_title(subTitles[ sub_count - 1] + ' %d'%len(matrix_by_dist[i]))
setTickAndLabel(fas, ticks, labels)
#fes.imshow(matrix_exist[matrix_exist_by_dist[i]], aspect='auto')
fes.plot(matrix_exist[matrix_exist_by_dist[i]].sum(axis=0)/matrix_exist[matrix_exist_by_dist[i]].shape[0])
fes.set_title(subTitles[ sub_count - 1] + ' %d'%len(matrix_exist_by_dist[i]))
setTickAndLabel(fes, ticks, labels)
sub_count += 1
else:
ncolumns = 1
NUMPLOTS=1
fas = fig_all.add_subplot((NUMPLOTS+1)/2, ncolumns, sub_count)
fes = fig_exist.add_subplot((NUMPLOTS+1)/2, ncolumns, sub_count)
matrix_all_sorted = matrix_all[matrix_all_idx]
matrix_exist_sorted = matrix_exist[matrix_exist_idx]
fas.plot(matrix_all_sorted.sum(axis=0)/matrix_all_sorted.shape[0])
fas.set_title("All dist %d" % len(matrix_all_idx))
setTickAndLabel(fas, ticks, labels)
fig_all.suptitle("Heatmap all for %s"%motifs[0])
#pl.colorbar()
#fig_all.close()
#pl.close()
fes.plot(matrix_exist_sorted.sum(axis=0)/matrix_exist_sorted.shape[0])
fig_exist.suptitle("Heatmap exist for %s"%motifs[0])
fes.set_title("All dist %d" % len(matrix_exist_idx))
setTickAndLabel(fes, ticks, labels)
fig_all.subplots_adjust(right=0.8)
fig_exist.subplots_adjust(right=0.8)
#cbar_ax = fig_all.add_axes([0.85,0.15,0.05,0.7])
#cbar_ax_e = fig_exist.add_axes([0.85,0.15,0.05,0.7])
#fig_all.colorbar(ima, cax=cbar_ax)
#fig_all.show()
#pl.show()
fig_all.tight_layout()
fig_exist.tight_layout()
saveOrPrint(fig_all, "Heatmap_all", '', savefig)
#fig_exist.colorbar(ime, cax=cbar_ax_e)
#fig_exist.show()
saveOrPrint(fig_exist, "Heatmap_exist", '', savefig)
