#!/usr/bin/env python
from splicesite import SpliceSite
import fruitfly
with open('sequence.txt','rb') as f:
a = f.read()
with open('variant.txt','rb') as f:
b = f.read()
sitesA = fruitfly.getSpliceSitePredictions(a)
sitesB = fruitfly.getSpliceSitePredictions(b)
print "~"*10 + " Expected " + "~"*10
for ss in sitesA:
print ss
print ""
print "~"*10 + " Variant " + "~"*10
for ss in sitesB:
print ss
print ""
print "~"*10 + " Comparing Results " + "~"*10
print ""
# List where we will store the SpliceSites to print
# These are the sites that are "interesting"
printList = []
def predictSpliceSites(self,
rows,
genomeFile,
db='hg38',
chromcol='#CHROMCOL',
poscol='POS',
varcol='VARIANT'):
# celery kung fu
self.predictions = list()
#self.warnings = list()
warnings = list()
task_id = predictSpliceSites.request.id
rowCount = len(rows)
for idx, row in enumerate(rows):
logger.info('Processing row %d' % (idx))
warnings.append('Processing row %d' % (idx))
chrom = "chr%s" % (row[chromcol])
seqStart = int(row[poscol]) - 501
seqEnd = int(row[poscol]) + 500
genome = sequenceutils.loadGenome(genomeFile)
seq = sequenceutils.getSequence(genome, chrom, seqStart, seqEnd)
# make a copy of seq but with the base modified at specified position
# python doesn't support item assignments w/in strings so build pieces
var = seq[0:500]
var = var + row[varcol]
var = var + seq[501:]
expectedSites = fruitfly.getSpliceSitePredictions(seq)
variantSites = fruitfly.getSpliceSitePredictions(var)
expectedList = []
variantList = []
msgList = []
# not used ... yet
for ss in expectedSites:
expectedList.append(
[ss.start, ss.end, ss.score, ss.intron, ss.exon])
for ss in variantSites:
variantList.append(
[ss.start, ss.end, ss.score, ss.intron, ss.exon])
for i, variantSite in enumerate(variantSites):
(posB, baseB, scoreB) = variantSite.getSpliceSite()
foundMatch = False
for j, expectedSite in enumerate(expectedSites):
(posA, baseA, scoreA) = expectedSite.getSpliceSite()
if posA == posB:
foundMatch = True
if scoreA == scoreB:
if baseA != baseB:
#flash("Base changed from %s to %s at position %d "
# "with score %0.2f\n"%(baseA, baseB, posA, scoreA))
warn = (
"Base changed from %s to %s at position %d "
"with score %0.2f" %
(baseA, baseB, posA, scoreA))
warnings.append(warn)
else:
delta = abs(scoreA - scoreB)
if delta >= 0.4:
#flash("Score changed by %0.2f from %0.2f to %0.2f "
# "at position %d.\n"%(delta, scoreA, scoreB, posA))
warn = (
"Score changed by %0.2f from %0.2f to %0.2f "
"at position %d." %
(delta, scoreA, scoreB, posA))
warnings.append(warn)
if not foundMatch:
if posB >= len(seq):
#flash("Oops! Somehow the length of our expected and varied "
# "sequences is not equivalent (%d for expected, "
# "%d for varied).\n"%(len(seq),len(var)))
warn = (
"Oops! Somehow the length of our expected and varied "
"sequences is not equivalent (%d for expected, "
"%d for varied)." % (len(seq), len(var)))
warnings.append(warn)
break
origBase = seq[posB].lower()
msgList.append(
"New splice site predicted at position %d with score %0.2f. "
"Original base was '%s' and new base is '%s'." %
(posB, scoreB, origBase, baseB))
# Add to the session predictions list
self.predictions.append(variantSite)
logger.debug('Creating output file ...')
filename = str(task_id) + '.csv'
path = os.path.join(app.config['UPLOAD_FOLDER'], filename)
logger.debug('Filename: %s' % (filename))
logger.debug('Path: %s' % (path))
fileutils.predictionsToCsv(self.predictions, path)
logger.debug('Done writing to file')
logger.info('Returning result')
return {'current': rowCount, 'total': rowCount, 'warnings': warnings}
def predictSpliceSites(self,
rows,
genomeFile,
db='hg38',
chromcol='#CHROMCOL',
poscol='POS',
varcol='VARIANT'):
# celery kung fu
self.predictions = list()
#self.warnings = list()
warnings = list()
task_id = predictSpliceSites.request.id
rowCount = len(rows)
for idx,row in enumerate(rows):
logger.info('Processing row %d'%(idx))
warnings.append('Processing row %d'%(idx))
chrom = "chr%s"%(row[chromcol])
seqStart = int(row[poscol]) - 501
seqEnd = int(row[poscol]) + 500
genome = sequenceutils.loadGenome(genomeFile)
seq = sequenceutils.getSequence(genome, chrom, seqStart, seqEnd)
# make a copy of seq but with the base modified at specified position
# python doesn't support item assignments w/in strings so build pieces
var = seq[0:500]
var = var + row[varcol]
var = var + seq[501:]
expectedSites = fruitfly.getSpliceSitePredictions(seq)
variantSites = fruitfly.getSpliceSitePredictions(var)
expectedList = []
variantList = []
msgList = []
# not used ... yet
for ss in expectedSites:
expectedList.append([ss.start,
ss.end,
ss.score,
ss.intron,
ss.exon])
for ss in variantSites:
variantList.append([ss.start,
ss.end,
ss.score,
ss.intron,
ss.exon])
for i, variantSite in enumerate(variantSites):
(posB,baseB,scoreB) = variantSite.getSpliceSite()
foundMatch = False
for j, expectedSite in enumerate(expectedSites):
(posA,baseA,scoreA) = expectedSite.getSpliceSite()
if posA==posB:
foundMatch = True
if scoreA==scoreB:
if baseA != baseB:
#flash("Base changed from %s to %s at position %d "
# "with score %0.2f\n"%(baseA, baseB, posA, scoreA))
warn = ("Base changed from %s to %s at position %d "
"with score %0.2f"%(baseA,baseB,posA,scoreA))
warnings.append(warn)
else:
delta = abs(scoreA - scoreB)
if delta >= 0.4:
#flash("Score changed by %0.2f from %0.2f to %0.2f "
# "at position %d.\n"%(delta, scoreA, scoreB, posA))
warn = ("Score changed by %0.2f from %0.2f to %0.2f "
"at position %d."%(delta,scoreA,scoreB,posA))
warnings.append(warn)
if not foundMatch:
if posB >= len(seq):
#flash("Oops! Somehow the length of our expected and varied "
# "sequences is not equivalent (%d for expected, "
# "%d for varied).\n"%(len(seq),len(var)))
warn = ("Oops! Somehow the length of our expected and varied "
"sequences is not equivalent (%d for expected, "
"%d for varied)."%(len(seq),len(var)))
warnings.append(warn)
break
origBase = seq[posB].lower()
msgList.append("New splice site predicted at position %d with score %0.2f. "
"Original base was '%s' and new base is '%s'."
%(posB, scoreB, origBase, baseB))
# Add to the session predictions list
self.predictions.append(variantSite)
logger.debug('Creating output file ...')
filename = str(task_id) + '.csv'
path = os.path.join(app.config['UPLOAD_FOLDER'],filename)
logger.debug('Filename: %s'%(filename))
logger.debug('Path: %s'%(path))
fileutils.predictionsToCsv(self.predictions,path)
logger.debug('Done writing to file')
logger.info('Returning result')
return {'current': rowCount, 'total': rowCount, 'warnings': warnings }
def splice_site():
if request.method == 'POST':
session['db'] = 'hg38' # use the latest genome by default
if 'db' in request.form:
if request.form['db'] == 'hg19':
session['db'] = 'hg19'
elif request.form['db'] != 'hg38':
flash('Unsuported genome option %s. Using hg38.' %
(request.form['session']))
else:
flash('Genome not specified. Using hg38.')
# Set up the mutation information for this run
session['chromosome'] = 'chr1'
session['position'] = '1'
session['base'] = 'A'
if 'chromosome' in request.form and \
request.form['chromosome'] != session['chromosome']:
session['chromosome'] = request.form['chromosome']
if 'position' in request.form and \
request.form['position'] != session['position']:
session['position'] = request.form['position']
if 'base' in request.form and \
request.form['base'] != session['base']:
session['base'] = request.form['base']
createSession()
genomePath = "genomes"
if session['db'] == 'hg19':
genomeFile = "hg19.2bit"
else:
session['db'] = 'hg38'
genomeFile = "hg38.2bit"
genomeFilePath = "/".join([genomePath, genomeFile])
#if app.config['GB'] == 'UCSC':
# seq = genomebrowser.gb_getSequence(hgsid, db=db, chrom=chrom,
# left=(int(pos)-1),
# right=(int(pos)),
# leftPad=500,
# rightPad=500)
#else:
chrom = "chr%s" % (session['chromosome'])
seqStart = int(session['position']) - 501
seqEnd = int(session['position']) + 500
genome = sequenceutils.loadGenome(genomeFilePath)
seq = sequenceutils.getSequence(genome, chrom, seqStart, seqEnd)
# make a copy of seq but with the base modified at specified position
# python doesn't support item assignments w/in strings so build pieces
var = seq[0:500]
var = var + session['base']
var = var + seq[501:]
expectedSites = fruitfly.getSpliceSitePredictions(seq)
variantSites = fruitfly.getSpliceSitePredictions(var)
expectedList = []
variantList = []
msgList = []
for ss in expectedSites:
expectedList.append(
[ss.start, ss.end, ss.score, ss.intron, ss.exon])
for ss in variantSites:
variantList.append(
[ss.start, ss.end, ss.score, ss.intron, ss.exon])
# List where we will store the SpliceSites to print
# These are the sites that are "interesting"
reportList = []
for i, variantSite in enumerate(variantSites):
(posB, baseB, scoreB) = variantSite.getSpliceSite()
foundMatch = False
for j, expectedSite in enumerate(expectedSites):
(posA, baseA, scoreA) = expectedSite.getSpliceSite()
if posA == posB:
foundMatch = True
if scoreA == scoreB:
if baseA != baseB:
flash("Base changed from %s to %s at position %d "
"with score %0.2f\n" %
(baseA, baseB, posA, scoreA))
else:
delta = abs(scoreA - scoreB)
if delta >= 0.4:
flash("Score changed by %0.2f from %0.2f to %0.2f "
"at position %d.\n" %
(delta, scoreA, scoreB, posA))
if not foundMatch:
if posB >= len(seq):
flash(
"Oops! Somehow the length of our expected and varied "
"sequences is not equivalent (%d for expected, "
"%d for varied).\n" % (len(seq), len(var)))
break
origBase = seq[posB].lower()
msgList.append(
"New splice site predicted at position %d with score %0.2f. "
"Original base was '%s' and new base is '%s'." %
(posB, scoreB, origBase, baseB))
reportList.append([
variantSite.start, variantSite.end, variantSite.score,
variantSite.intron, variantSite.exon
])
return render_template('splicesite.html',
expectedList=expectedList,
variantList=variantList,
msgList=msgList,
reportList=reportList,
db=session['db'],
chromosome=session['chromosome'],
position=session['position'],
base=session['base'])
# if not post, return index.html
return render_template('index.html')
def splice_site():
if request.method == 'POST':
session['db'] = 'hg38' # use the latest genome by default
if 'db' in request.form:
if request.form['db'] == 'hg19':
session['db'] = 'hg19'
elif request.form['db'] != 'hg38':
flash('Unsuported genome option %s. Using hg38.'
%(request.form['session']))
else:
flash('Genome not specified. Using hg38.')
# Set up the mutation information for this run
session['chromosome'] = 'chr1'
session['position'] = '1'
session['base'] = 'A'
if 'chromosome' in request.form and \
request.form['chromosome'] != session['chromosome']:
session['chromosome'] = request.form['chromosome']
if 'position' in request.form and \
request.form['position'] != session['position']:
session['position'] = request.form['position']
if 'base' in request.form and \
request.form['base'] != session['base']:
session['base'] = request.form['base']
createSession()
genomePath = "genomes"
if session['db'] == 'hg19':
genomeFile = "hg19.2bit"
else:
session['db'] = 'hg38'
genomeFile = "hg38.2bit"
genomeFilePath = "/".join([genomePath,genomeFile])
#if app.config['GB'] == 'UCSC':
# seq = genomebrowser.gb_getSequence(hgsid, db=db, chrom=chrom,
# left=(int(pos)-1),
# right=(int(pos)),
# leftPad=500,
# rightPad=500)
#else:
chrom = "chr%s"%(session['chromosome'])
seqStart = int(session['position']) - 501
seqEnd = int(session['position']) + 500
genome = sequenceutils.loadGenome(genomeFilePath)
seq = sequenceutils.getSequence(genome, chrom, seqStart, seqEnd)
# make a copy of seq but with the base modified at specified position
# python doesn't support item assignments w/in strings so build pieces
var = seq[0:500]
var = var + session['base']
var = var + seq[501:]
expectedSites = fruitfly.getSpliceSitePredictions(seq)
variantSites = fruitfly.getSpliceSitePredictions(var)
expectedList = []
variantList = []
msgList = []
for ss in expectedSites:
expectedList.append([ss.start,
ss.end,
ss.score,
ss.intron,
ss.exon])
for ss in variantSites:
variantList.append([ss.start,
ss.end,
ss.score,
ss.intron,
ss.exon])
# List where we will store the SpliceSites to print
# These are the sites that are "interesting"
reportList = []
for i, variantSite in enumerate(variantSites):
(posB,baseB,scoreB) = variantSite.getSpliceSite()
foundMatch = False
for j, expectedSite in enumerate(expectedSites):
(posA,baseA,scoreA) = expectedSite.getSpliceSite()
if posA==posB:
foundMatch = True
if scoreA==scoreB:
if baseA != baseB:
flash("Base changed from %s to %s at position %d "
"with score %0.2f\n"%(baseA, baseB, posA, scoreA))
else:
delta = abs(scoreA - scoreB)
if delta >= 0.4:
flash("Score changed by %0.2f from %0.2f to %0.2f "
"at position %d.\n"%(delta, scoreA, scoreB, posA))
if not foundMatch:
if posB >= len(seq):
flash("Oops! Somehow the length of our expected and varied "
"sequences is not equivalent (%d for expected, "
"%d for varied).\n"%(len(seq),len(var)))
break
origBase = seq[posB].lower()
msgList.append("New splice site predicted at position %d with score %0.2f. "
"Original base was '%s' and new base is '%s'."
%(posB, scoreB, origBase, baseB))
reportList.append([variantSite.start,
variantSite.end,
variantSite.score,
variantSite.intron,
variantSite.exon])
return render_template('splicesite.html',
expectedList=expectedList,
variantList=variantList,
msgList=msgList,
reportList=reportList,
db=session['db'],
chromosome=session['chromosome'],
position=session['position'],
base=session['base'])
# if not post, return index.html
return render_template('index.html')