def cns_opp_strand(cnss, qgene, sgene):
cnss = list(cnss)
cnss = map(change_orient,cnss)
sgene[0] *= -1
sgene[1] *= -1
cnss = [(c[0], c[1], c[2], c[3],c[-2]) for c in remove_crossing_cnss(cnss, qgene, sgene)]
cnss_fixed = [(c[0], c[1], -c[2], -c[3],c[-1]) for c in cnss]
return cnss_fixed
def remove_crossing_hits(exon_hits,qfeat,sfeat):
"""uses find cns remove overlaping hits and corssing on each exon"""
qgene =[qfeat['start'], qfeat['end']]
sgene =[sfeat['start'], sfeat['end']]
orient = qfeat['strand'] == sfeat['strand'] and 1 or -1
exon_hits = list(exon_hits)
if orient == -1:
for i, hit in enumerate(exon_hits):
hit = list(hit)
hit[2] *= -1
hit[3] *= -1
exon_hits[i] = tuple(hit)
sgene[0] *= -1
sgene[1] *= -1
non_crossing_hits = [(c[0], c[1], c[2], c[3], c[-2]) for c in remove_crossing_cnss(exon_hits,qgene,sgene)]
if orient == -1:
non_crossing_hits == [(c[0],c[1],-c[2], -c[3], c[-1]) for c in remove_crossing_cnss(exon_hits,qgene,sgene)]
#non_crossing_dict = {str(locs):psudo[str(locs)] for locs in non_crossing_hits}
return non_crossing_hits
def protein_parse(hit,gene,gene_bed, hit_bed):
"creates a protein fasta and non translated exon fasta \
blastx them and parse the results"
hit_fasta = "{0}q.fasta".format('/Users/gturco/code/freeling_lab/pseudo/data/rice_v6_setaria64/')
gene_fasta = "{0}s.fasta".format('/Users/gturco/code/freeling_lab/pseudo/data/rice_v6_setaria64/')
if len(re.findall('X',gene_bed.row_cds_sequence(gene['accn']))) >0:
return "masked", "masked","masked"
else:
protein_fasta(hit_bed,hit,False,hit_fasta)
protein_fasta(gene_bed,gene,True,gene_fasta)
#cmd = "/Users/gturco/blast-2.2.25/bin/bl2seq -p blastx -G 11 -E 1 -W 3 -e 0.001 -D 1 -i {0} -j {1} | grep -v '#' | grep -v 'WARNING' | grep -v 'ERROR'".format(hit_fasta,gene_fasta)
cmd = "/Users/gturco/ncbi-blast-2.2.25+/bin/blastx -gapopen 11 -gapextend 1 -word_size 3 -evalue 0.001 -outfmt '7 gaps qframe std' -query {0} -subject {1} | grep -v '#' | grep -v 'WARNING' | grep -v 'ERROR'".format(hit_fasta,gene_fasta)
#print >>sys.stderr, "{1} {2} cmd : {0} ".format(cmd,gene,hit)
res = commands.getoutput(cmd)
print >>sys.stderr, res
frame_dict = {'1':{"alignment":[],"qstart":[],"gaps":[]},
'2':{"alignment":[],"qstart":[],"gaps":[]},
'3':{"alignment":[],"qstart":[],"gaps":[]},'-1':{"alignment":[],"qstart":[],"gaps":[]},
'-2':{"alignment":[],"qstart":[],"gaps":[]},
'-3':{"alignment":[],"qstart":[],"gaps":[]}}
qhit =[hit['start'], hit['end']]
sgene =[gene['start'], gene['end']]
locs_list = []
for line in res.split("\n"):
if not line: continue
if "WARNING:" in line: continue
if "ERROR" in line: continue
line = line.split("\t")
locs = map(int, line[8:12])
locs.extend(map(float, line[12:]))
frame = line[1]
gaps = line[0]
qstart = min(locs[0],locs[1])
length = line[5]
frame_dict[frame]["alignment"].append(length)
frame_dict[frame]["qstart"].append(qstart)
frame_dict[frame]["gaps"].append(gaps)
#frame_lengths = [(sum(frame_dict[key]),key) for key in frame_dict.keys()]
#frame_lengths.sort()
#largest_frame = frame_lengths[-1][1]
#if largest_frame < len....:
# frame_shift
#find stop codon from largest frame + start site....
locs = tuple(locs)
locs_list.update((locs,))
#print >>sys.stderr, "locs_list: {0}".format(locs_list)
#non_crossing = [(c[0], c[1], c[2], c[3]) for c in remove_intersecting_hits(list(locs_list))]
non_crossing = [(c[0], c[1], c[2], c[3], c[4]) for c in remove_crossing_cnss(list(locs_list),qhit,sgene)]
frame_shift = False
if len(non_crossing) > 1:
frame_shift = False
total_hit_len = sum([abs(q_start-q_end) for q_start,q_end,s_start,s_end, evalu in non_crossing])
total_gene_len = sum(abs(s_start-s_end) for q_start,q_end, s_start,s_end, evalu in non_crossing)
print >>sys.stderr,non_crossing
ref_hit_len = len(hit_bed.row_cds_sequence(hit['accn']))
ref_gene_len = len(gene_bed.row_cds_sequence(gene['accn']))
print >>sys.stderr,"hit_total {0} \n gene_len {1}".format(total_hit_len,ref_hit_len)
#print >>sys.stderr, total_hit_len
hit_len = total_hit_len/float(ref_hit_len)
gene_len = total_gene_len/float(ref_gene_len/3)
return hit_len, gene_len, frame_shift
def parse_blast(blast_str, orient, qfeat, sfeat, qbed, sbed, qpad, spad, unmasked_fasta):
blast = []
slope = orient
qgene = [qfeat['start'], qfeat['end']]
sgene = [sfeat['start'], sfeat['end']]
sgene = sgene[::slope]
center = sum(qgene)/2., sum(sgene)/2.
intercept = center[1] - slope * center[0]
x = np.linspace(qgene[0] - qpad, qgene[1] + qpad, 50)
y = slope * x + intercept
feats_nearby = get_feats_nearby(qgene,sgene,qfeat,sfeat,x,y,qbed,sbed)
qgene_space_poly,qgene_poly,sgene_space_poly,sgene_poly = get_genespace(qfeat,sfeat,qgene,sgene)
intronic_removed = 0
cnss = set([])
for line in blast_str.split("\n"):
if "WARNING:" in line: continue
if "ERROR" in line: continue
if line == '': continue
line = line.split("\t")
if float(line[-1]) < 29.5: continue #finds 15/15 match
# if float(line[-1]) < 33.4: continue #finds 17/17 match
locs = map(int, line[6:10])
locs.extend(map(float, line[10:]))
xx = locs[:2]
yy = locs[2:4]
#######################################################
# MAIZE BOWTIE : JUST 5 PRIME 3 PRIME
#######################################################
qcenter = sum(qgene)/2
scenter = sum(sgene)/2 * orient
qcns_center = sum(xx)/2
scns_center = sum(yy)/2 * orient
if scns_center > scenter and qcns_center < qcenter: continue
if qcns_center > qcns_center and scns_center < scenter : continue
# to be saved. a hit must either be in an intron in both
# genes, or in neither.
##########################################################
# DEAL WITH INTRONIC cnss in the gene of interest.
##########################################################
xls = LineString([(0, locs[0]), (0, locs[1])])
yls = LineString([(0, locs[2]), (0, locs[3])])
locs = tuple(locs) # make it hashable.
if qgene_poly.intersects(xls) and sgene_poly.intersects(yls):
cnss.update((locs,))
continue
# has to be both or neither.
if qgene_space_poly.intersects(xls) or sgene_space_poly.intersects(yls):
intronic_removed += 1
continue
##########################################################
###############################################################
# for all other genes, if it's in an intron, we dont keep it.
###############################################################
intronic = False
# get rid of stuff that overlaps another gene:
for sub, (start, stop) in (('q', locs[:2]), ('s', locs[2:4])):
feats = feats_nearby[sub]
if feats is None: continue
# the current hsp is overlapping another gene. we dont want that...
if feats.contains(Point(0, start)) or feats.contains(Point(0, stop)):
intronic = True
break
if intronic: continue
##########################################################
cnss.update((locs,))
# cant cross with < 2 cnss.
# get rid of the eval, bitscore stuff.
if len(cnss) < 2: return [(c[0], c[1], c[2], c[3],c[-1]) for c in cnss]
cnss = list(cnss)
####################################################################################
#########split cns into groups based on inversion, seq marks in maize ##########
#################################################################################
def group_cns(cnss, group):
"""input list of cns and list of groups , this puts the cns in a dictionary fmt key = group
values = cns that fall within range of group"""
for cns in cnss:
if cns[2] in range(group[0],group[1]): # group start and end pos
key = group
cns_groups.setdefault(key, []).append(cns)
cns_groups = {}
inversion_groups = find_inversions(unmasked_fasta, sfeat, spad)
[group_cns(cnss, group) for group in inversion_groups] # creates dict where key = group value is appended cns
# for each goup of cns values run the followiung
cns_by_group = []
for key in cns_groups.keys():
# # first group, groups into smaller groups on strand
values = cns_groups[key]
opp_strand = []
same_strand = []
for cns in values:
if slope == 1 and cns[2] > cns[3]:
opp_strand.append(cns)
elif slope == -1 and cns[2] < cns[3]:
opp_strand.append(cns)
else:
same_strand.append(cns)
# need to flip to negative so the overlapping stuff still works.
if orient == -1:
same_strand = map(change_orient, same_strand)
opp_strand = map(change_orient, opp_strand)
sgene[0] *= -1
sgene[1] *= -1
if abs(sgene[1]) in range(key[0], key[1]): # if the cns fall in same group as gene we know its same stand as gene and dont need to run rest
cnss_same_strand = [(c[0], c[1], c[2], c[3],c[-1]) for c in remove_crossing_cnss(same_strand, qgene, sgene)]
map(cns_by_group.append, cnss_same_strand)
else:
cnss_same_strand = [(c[0], c[1], c[2], c[3],c[-1]) for c in remove_crossing_cnss(same_strand, qgene, sgene)]
cnss_opp_strand = cns_opp_strand(opp_strand, qgene, sgene) # alternitive for cns on opp strand
if len(cnss_same_strand) < len(cnss_opp_strand):
map(cns_by_group.append, cnss_opp_strand)
else: # what about if they are the same, use non reverse complment
map(cns_by_group.append, cnss_same_strand)
if orient == -1:
cns_by_group = [(c[0], c[1], -c[2], -c[3],c[-1]) for c in cns_by_group]
return cns_by_group