def calc_stats(mfiles_pat, gff):
fat = Fat(gff)
hists = {}
for i in range(1, 6):
i = str(i)
mf = mfiles_pat % i
assert os.path.exists(mf)
hists[i] = np.fromfile(mf, dtype=np.float32)
header = "accn,gene,cds,intron,up10,up100,up1000,down10,down100,down1000"
print header
for accn, f in sorted(fat.iteritems()):
data = [accn]
if not f.seqid in hists: continue # C, G chrs
hist = hists[f.seqid]
for locs in ([[f.start, f.end]], getattr(f, 'CDS',
None), fat.introns(f),
fat.upstream(f, 10, noncoding=True),
fat.upstream(f, 100, noncoding=True),
fat.upstream(f, 1000, noncoding=True),
fat.downstream(f, 10, noncoding=True),
fat.downstream(f, 100, noncoding=True),
fat.downstream(f, 1000, noncoding=True)):
if locs is None:
# occurs when there's no CDS.
data.append("na")
continue
slicer = pairs_to_slice(locs)
try:
m = hist[slicer] # this context.
except IndexError:
# difference between fasta and features due to version
slicer = slicer[slicer < hist.shape[0]]
m = hist[slicer]
data.append("%.5f" % m.mean())
print ",".join(data)
def calc_stats(mfiles_pat, gff):
fat = Fat(gff)
hists = {}
for i in range(1, 6):
i = str(i)
mf = mfiles_pat % i
assert os.path.exists(mf)
hists[i] = np.fromfile(mf, dtype=np.float32)
header = "accn,gene,cds,intron,up10,up100,up1000,down10,down100,down1000"
print header
for accn, f in sorted(fat.iteritems()):
data = [accn]
if not f.seqid in hists: continue # C, G chrs
hist = hists[f.seqid]
for locs in ([[f.start, f.end]],
getattr(f, 'CDS', None),
fat.introns(f),
fat.upstream(f, 10, noncoding=True),
fat.upstream(f, 100, noncoding=True),
fat.upstream(f, 1000, noncoding=True),
fat.downstream(f, 10, noncoding=True),
fat.downstream(f, 100, noncoding=True),
fat.downstream(f, 1000, noncoding=True)
):
if locs is None:
# occurs when there's no CDS.
data.append("na")
continue
slicer = pairs_to_slice(locs)
try:
m = hist[slicer] # this context.
except IndexError:
# difference between fasta and features due to version
slicer = slicer[slicer < hist.shape[0]]
m = hist[slicer]
data.append("%.5f" % m.mean())
print ",".join(data)
def calc_stats(mfiles_pat, gff):
fat = Fat(gff)
methyl = {}
contexts = {}
for i in range(1, 6):
i = str(i)
mf = mfiles_pat % i
assert os.path.exists(mf)
methyl[i] = np.fromfile(mf, dtype=np.float32)
mt = np.fromfile(mf.replace(".methyl.", ".methyltype."), dtype=np.uint8)
# these can be used to mask to a given context.
contexts[i] = {'cg': (mt == 1) | (mt == 4),
'chg': (mt == 2) | (mt == 5),
'chh': (mt == 3) | (mt == 6)}
header = ["accn"]
for ctx in ('cg', 'chg', 'chh'):
# TODO: make this suck less.
header.append(",".join([
"gene_CTX_avg,gene_CTX_avg_gt0,gene_CTX_n_methylated,gene_CTX_n,gene_CTX_%_methylated",
"cds_CTX_avg,cds_CTX_avg_gt0,cds_CTX_n_methylated,cds_CTX_n,cds_CTX_%_methylated",
"intron_CTX_avg,intron_CTX_avg_gt0,intron_CTX_n_methylated,intron_CTX_n,intron_CTX_%_methylated",
"up10_CTX_avg,up10_CTX_avg_gt0,up10_CTX_n_methylated,up10_CTX_n,up10_CTX_%_methylated",
"up100_CTX_avg,up100_CTX_avg_gt0,up100_CTX_n_methylated,up100_CTX_n,up100_CTX_%_methylated",
"up1000_CTX_avg,up1000_CTX_avg_gt0,up1000_CTX_n_methylated,up1000_CTX_n,up1000_CTX_%_methylated",
"down10_CTX_avg,down10_CTX_avg_gt0,down10_CTX_n_methylated,down10_CTX_n,down10_CTX_%_methylated",
"down100_CTX_avg,down100_CTX_avg_gt0,down100_CTX_n_methylated,down100_CTX_n,down100_CTX_%_methylated",
"down1000_CTX_avg,down1000_CTX_avg_gt0,down1000_CTX_n_methylated,down1000_CTX_n,down1000_CTX_%_methylated"
]).replace('CTX', ctx)) # bleckh. shrug.
print ",".join(header)
for accn, f in sorted(fat.iteritems()):
data = [accn]
if not f.seqid in contexts: continue # C, G
for mtype, context in sorted(contexts[f.seqid].iteritems()):
for locs in ([[f.start, f.end]],
getattr(f, 'CDS', None),
fat.introns(f),
fat.upstream(f, 10, noncoding=True),
fat.upstream(f, 100, noncoding=True),
fat.upstream(f, 1000, noncoding=True),
fat.downstream(f, 10, noncoding=True),
fat.downstream(f, 100, noncoding=True),
fat.downstream(f, 1000, noncoding=True)
):
if locs is None:
# occurs when there's no CDS.
data.extend(["na","na","na","na","na"])
continue
slicer = pairs_to_slice(locs)
try:
ctx = context[slicer] # this context.
except IndexError:
# difference between fasta and features due to version
slicer = slicer[slicer < context.shape[0]]
ctx = context[slicer]
# methylation for this CDS masked to current context
me = methyl[f.seqid][slicer] * ctx
# number of sites in this context.
ctx_sites = ctx.sum()
# number of site in this context that are methylated.
m_ctx_sites = (me > 0).sum()
# proprtion of sites that can be methylated that are.
p_methylated = float(m_ctx_sites) / ctx_sites
# average methylation for sites in this context.
avg_methyl = me.sum() / float(ctx_sites)
# average methylation for sites that are methylated. (exclude zeros).
avg_methyl_gt0 = me.sum() / float(m_ctx_sites)
data.extend(["%.5f" % d for d in [avg_methyl, avg_methyl_gt0]])
data.extend(["%i" % d for d in [m_ctx_sites, ctx_sites]])
data.append("%.5f" % p_methylated)
print ",".join(data)