def spatial_plot_plus_links(bead_xy_a,
bead_pairs,
umi_dist,
title: str,
pdf_pages: PdfPages,
pct: float = 95):
with new_ax(pdf_pages, include_fig=True) as (fig, ax):
# version of 'Blues' colormap that is pure white at the bottom
cmap = matplotlib.colors.LinearSegmentedColormap.from_list(
"BluesW", [(1.0, 1.0, 1.0), (0.0314, 0.188, 0.450)])
norm = matplotlib.colors.Normalize(0,
np.percentile(umi_dist, pct),
clip=True)
c = ax.scatter(
bead_xy_a[:, 0],
bead_xy_a[:, 1],
c=umi_dist,
s=0.5,
cmap=cmap,
norm=norm,
)
c.set_rasterized(True)
xs = bead_xy_a[bead_pairs, 0].T
ys = bead_xy_a[bead_pairs, 1].T
ax.plot(xs, ys, alpha=0.1, color="g", linewidth=0.1)
ax.set_xlabel("X")
ax.set_ylabel("Y")
ax.axis("equal")
ax.set_title(title)
fig.colorbar(c, ax=ax)
def plot_base_distribution(pdf_pages: PdfPages, base_dist_file: Path,
title: str):
with base_dist_file.open() as fh:
rows = list(csv.reader(fh, delimiter="\t"))
base_distribution = np.array([[float(v) for v in r[1:-1]]
for r in rows[1:]])
base_distribution /= base_distribution.sum(axis=1, keepdims=True)
with new_ax(pdf_pages) as ax:
ax.set_prop_cycle("color", ["red", "blue", "green", "purple"])
ax.plot(
np.arange(1, base_distribution.shape[0] + 1),
base_distribution,
linewidth=0,
marker="o",
markersize=10,
alpha=0.8,
label=["A", "C", "G", "T"],
)
ax.legend(loc="lower right")
ax.set_xlim(0, base_distribution.shape[0] + 1)
ax.set_ylim(0, np.max(base_distribution) + 0.02)
ax.set_xlabel("base position")
ax.set_ylabel("fraction of reads")
ax.set_title(title)
def plot_poly_a_trimming(pdf_pages: PdfPages, qm: dict[str, int],
poly_a_summary_files: list[Path]):
total_hist = Counter()
for summary_file in poly_a_summary_files:
_, poly_a_hist = read_dropseq_metrics(summary_file)
total_hist += poly_a_hist
trimmed_count = sum(total_hist[k] for k in total_hist if k > 0)
with new_ax(pdf_pages) as ax:
ax.plot(
np.arange(1,
max(total_hist) + 1),
[total_hist[k] for k in total_hist if k > 0],
marker="o",
color="k",
)
ax.set_xlabel("First base of PolyA tail trimmed")
ax.set_ylabel("Number of reads")
ax.set_title(
f"% Reads trimmed by 3' PolyA trimmer: {trimmed_count / qm['Total']:.3%}"
)
ax.set_xlim(0, max(total_hist) + 1)
def plot_log_hist(dist, title: str, pdf_pages: PdfPages):
max_d = np.ceil(np.log10(max(dist)))
with new_ax(pdf_pages) as ax:
ax.hist(dist,
bins=np.logspace(0, max_d, int(max_d * 10 + 1)),
log=True)
ax.set_xscale("log")
ax.set_title(title)
def plot_combined_mapping_quality(pdf_pages: PdfPages, quality_metrics: Path,
matched_quality_metrics: Path):
keys = ["Total", "Mapped", "HQ", "HQ No Dupes"]
qm, _ = read_quality_metrics(quality_metrics)
mm, _ = read_quality_metrics(matched_quality_metrics)
ms = [qm, mm]
x = np.arange(len(keys))
with new_ax(pdf_pages, ax_bounds=(0.1, 0.2, 0.8, 0.7)) as ax:
ax.set_prop_cycle("color", ["lightskyblue", "goldenrod"])
for i, m in enumerate(ms):
ax.bar(
x + 0.1 + 0.4 * i,
[m[k] for k in keys],
width=0.4,
align="edge",
label=["All", "Matched"][i],
edgecolor="black",
)
ax.set_xticks(x + 0.5)
ax.set_xticklabels(keys)
ax.set_xticks(
[v + 0.3 + i for v in x for i in (0.0, 0.4)],
minor=True,
)
ax.set_xticklabels(
[
f"{int(m[k]):,}\n({m[k] / m['Total']:.1%})" for k in keys
for m in ms
],
minor=True,
rotation=90,
)
ax.set_ylabel("# Reads")
ax.set_title("Alignment Quality")
ax.tick_params(axis="x", which="major", bottom=False, length=80)
ax.tick_params(axis="x", which="minor")
ax.legend()
return qm
def plot_reads_per_barcode(pdf_pages: PdfPages, barcode_count):
with gzip.open(barcode_count, "rt") as fh:
# this file has form `num_reads barcodes` in descending order
read_counts = [
int(r[0]) for r in csv.reader(fh, delimiter="\t") if r[0][0] != "#"
]
total_reads = sum(read_counts)
# truncate to first 10% of barcodes
read_counts = read_counts[:len(read_counts) // 10]
with new_ax(pdf_pages) as ax:
ax.plot(np.cumsum(read_counts) / total_reads, color="g")
ax.set_ylim((0, 1))
ax.set_xlabel("Top 10% of cell barcodes, by number of reads")
ax.set_ylabel("Cumulative fraction of reads")
ax.set_title("Cumulative fraction of reads per cell barcode")
def plot_spatial_distribution(pdf_pages: PdfPages, bead_xy: np.ndarray,
dist: list[float], title: str):
with new_ax(pdf_pages, include_fig=True) as (fig, ax):
c = ax.scatter(
bead_xy[:, 0],
bead_xy[:, 1],
c=dist,
s=0.5,
cmap="viridis_r",
norm=matplotlib.colors.Normalize(0,
np.percentile(dist, 95),
clip=True),
)
c.set_rasterized(True)
ax.set_xlabel("X")
ax.set_ylabel("Y")
ax.axis("equal")
ax.set_title(f"{title} per matched bead")
fig.colorbar(c, ax=ax)
def plot_mapping_quality(pdf_pages: PdfPages, quality_metrics: Path):
keys = ["Total", "Mapped", "HQ", "HQ No Dupes"]
qm, _ = read_quality_metrics(quality_metrics)
with new_ax(pdf_pages) as ax:
ax.bar(
[
f"{k}\n{int(qm[k]):,}\n({qm[k] / qm['Total']:.1%})"
for k in keys
],
[qm[k] for k in keys],
width=0.8,
color="lightskyblue",
edgecolor="black",
)
ax.set_ylabel("# Reads")
ax.set_title("Alignment Quality for All Reads")
return qm
def spatial_plot(bead_xy_a, dist, title, pdf_pages, pct: float = 95.0):
with new_ax(pdf_pages, include_fig=True) as (fig, ax):
# version of 'Blues' colormap that is pure white at the bottom
cmap = matplotlib.colors.LinearSegmentedColormap.from_list(
"BluesW", [(1.0, 1.0, 1.0), (0.0314, 0.188, 0.450)]
)
c = ax.scatter(
bead_xy_a[:, 0],
bead_xy_a[:, 1],
c=dist,
s=0.5,
cmap=cmap,
norm=matplotlib.colors.Normalize(0, np.percentile(dist, pct), clip=True),
)
c.set_rasterized(True)
ax.set_xlabel("X")
ax.set_ylabel("Y")
ax.axis("equal")
ax.set_title(title)
fig.colorbar(c, ax=ax)
def plot_frac_intronic_exonic(pdf_pages: PdfPages, frac_intron_exon: Path,
title: str):
# plot the fraction of reads mapping to different regions
frac_ie, _ = read_frac_intronic_exonic(frac_intron_exon)
# calculate some derivative measures
frac_ie["exonic"] = frac_ie["coding"] + frac_ie["utr"]
frac_ie["genic"] = frac_ie["exonic"] + frac_ie["intronic"]
keys = ["ribosomal", "exonic", "genic", "intronic", "intergenic"]
with new_ax(pdf_pages) as ax:
ax.bar(
[f"{k}\n{frac_ie[k]:.1%}" for k in keys],
[100 * frac_ie[k] for k in keys],
width=0.7,
color="lightskyblue",
edgecolor="black",
)
ax.set_ylim(0, 100)
ax.set_ylabel("Percentage")
ax.set_title(title)
def plot_dge_summary(pdf_pages: PdfPages, summary_file: Path, bead_xy: BeadXY):
"""
Plot histograms of UMIs and genes per barcode
"""
barcodes, umis_per_bc, genes_per_bc = read_dge_summary(summary_file)
bead_xy = np.vstack([bead_xy[bc] for bc in barcodes])
for dist, title in ((umis_per_bc, "UMIs"), (genes_per_bc, "genes")):
with new_ax(pdf_pages) as ax:
# can safely assume no zeros in this distribution
# 10000 should be good enough for anyone
ax.hist(
dist,
bins=np.logspace(0, 4, 41),
facecolor="lightskyblue",
edgecolor="black",
)
ax.set_xscale("log")
ax.set_xlabel(f"Number of {title} (log10)")
ax.set_title(f"Histogram of {title} per matched barcode")
plot_spatial_distribution(pdf_pages, bead_xy, umis_per_bc, "UMIs")
def plot_hist(dist, title: str, pdf_pages: PdfPages):
with new_ax(pdf_pages) as ax:
ax.hist(dist, bins=100, log=True)
ax.set_title(title)