def gcdepth(args):
"""
%prog gcdepth sample_name tag
Plot GC content vs depth vs genomnic bins. Inputs are mosdepth output:
- NA12878_S1.mosdepth.global.dist.txt
- NA12878_S1.mosdepth.region.dist.txt
- NA12878_S1.regions.bed.gz
- NA12878_S1.regions.bed.gz.csi
- NA12878_S1.regions.gc.bed.gz
A sample mosdepth.sh script might look like:
```
#!/bin/bash
LD_LIBRARY_PATH=mosdepth/htslib/ mosdepth/mosdepth $1 \\
bams/$1.bam -t 4 -c chr1 -n --by 1000
bedtools nuc -fi GRCh38/WholeGenomeFasta/genome.fa \\
-bed $1.regions.bed.gz \\
| pigz -c > $1.regions.gc.bed.gz
```
"""
import hashlib
from jcvi.algorithms.formula import MAD_interval as confidence_interval
from jcvi.graphics.base import latex, plt, savefig, set2
p = OptionParser(gcdepth.__doc__)
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
sample_name, tag = args
# The tag is used to add to title, also provide a random (hashed) color
coloridx = int(hashlib.sha256(tag).hexdigest(), 16) % len(set2)
color = set2[coloridx]
# mosdepth outputs a table that we can use to plot relationship
gcbedgz = sample_name + ".regions.gc.bed.gz"
df = pd.read_csv(gcbedgz, delimiter="\t")
mf = df.loc[:, ("4_usercol", "6_pct_gc")]
mf.columns = ["depth", "gc"]
# We discard any bins that are gaps
mf = mf[(mf["depth"] > 0.001) | (mf["gc"] > 0.001)]
# Create GC bins
gcbins = defaultdict(list)
for i, row in mf.iterrows():
gcp = int(round(row["gc"] * 100))
gcbins[gcp].append(row["depth"])
gcd = sorted(
(k * 0.01, confidence_interval(v)) for (k, v) in gcbins.items())
gcd_x, gcd_y = zip(*gcd)
m, lo, hi = zip(*gcd_y)
# Plot
plt.plot(
mf["gc"],
mf["depth"],
".",
color="lightslategray",
ms=2,
mec="lightslategray",
alpha=0.1,
)
patch = plt.fill_between(
gcd_x,
lo,
hi,
facecolor=color,
alpha=0.25,
zorder=10,
linewidth=0.0,
label="Median +/- MAD band",
)
plt.plot(gcd_x, m, "-", color=color, lw=2, zorder=20)
ax = plt.gca()
ax.legend(handles=[patch], loc="best")
ax.set_xlim(0, 1)
ax.set_ylim(0, 100)
ax.set_title("{} ({})".format(latex(sample_name), tag))
ax.set_xlabel("GC content")
ax.set_ylabel("Depth")
savefig(sample_name + ".gcdepth.png")
def gcdepth(args):
"""
%prog gcdepth sample_name tag
Plot GC content vs depth vs genomnic bins. Inputs are mosdepth output:
- NA12878_S1.mosdepth.global.dist.txt
- NA12878_S1.mosdepth.region.dist.txt
- NA12878_S1.regions.bed.gz
- NA12878_S1.regions.bed.gz.csi
- NA12878_S1.regions.gc.bed.gz
A sample mosdepth.sh script might look like:
```
#!/bin/bash
LD_LIBRARY_PATH=mosdepth/htslib/ mosdepth/mosdepth $1 \\
bams/$1.bam -t 4 -c chr1 -n --by 1000
bedtools nuc -fi GRCh38/WholeGenomeFasta/genome.fa \\
-bed $1.regions.bed.gz \\
| pigz -c > $1.regions.gc.bed.gz
```
"""
import hashlib
from jcvi.algorithms.formula import MAD_interval as confidence_interval
from jcvi.graphics.base import latex, plt, savefig, set2
p = OptionParser(gcdepth.__doc__)
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
sample_name, tag = args
# The tag is used to add to title, also provide a random (hashed) color
coloridx = int(hashlib.sha1(tag).hexdigest(), 16) % len(set2)
color = set2[coloridx]
# mosdepth outputs a table that we can use to plot relationship
gcbedgz = sample_name + ".regions.gc.bed.gz"
df = pd.read_csv(gcbedgz, delimiter="\t")
mf = df.loc[:, ("4_usercol", "6_pct_gc")]
mf.columns = ["depth", "gc"]
# We discard any bins that are gaps
mf = mf[(mf["depth"] > .001) | (mf["gc"] > .001)]
# Create GC bins
gcbins = defaultdict(list)
for i, row in mf.iterrows():
gcp = int(round(row["gc"] * 100))
gcbins[gcp].append(row["depth"])
gcd = sorted((k * .01, confidence_interval(v))
for (k, v) in gcbins.items())
gcd_x, gcd_y = zip(*gcd)
m, lo, hi = zip(*gcd_y)
# Plot
plt.plot(mf["gc"], mf["depth"], ".", color="lightslategray", ms=2,
mec="lightslategray", alpha=.1)
patch = plt.fill_between(gcd_x, lo, hi,
facecolor=color, alpha=.25, zorder=10,
linewidth=0.0, label="Median +/- MAD band")
plt.plot(gcd_x, m, "-", color=color, lw=2, zorder=20)
ax = plt.gca()
ax.legend(handles=[patch], loc="best")
ax.set_xlim(0, 1)
ax.set_ylim(0, 100)
ax.set_title("{} ({})".format(latex(sample_name), tag))
ax.set_xlabel("GC content")
ax.set_ylabel("Depth")
savefig(sample_name + ".gcdepth.png")