Skip to content

reference.md

Latest commit

 

History

History
118 lines (92 loc) · 6.16 KB

reference.md

File metadata and controls

118 lines (92 loc) · 6.16 KB
Raw

Reference

region_bed

Headerless, tab-separated file. Read by GenomeAxis (via read_bed); the whitelist regions the genome axis is drawn from.

Column Required Type Description
#CHR Yes str Chromosome.
START Yes int Region start (0-based).
END Yes int Region end.
name No str 4th column (e.g. arm id); read but not used.

chrom_sizes

Headerless, whitespace-separated file. Read by GenomeAxis.

Column Required Type Description
chromosome Yes str Chromosome name. Row order sets the axis chromosome order.
length Yes int Chromosome length.

seg_df (HATCHet-style seg.ucn profile)

Integer copy-number profile, one row per segment. Consumed by plot_cnv_profile and, in the 2D form below, by plot_scatter_2d.

Column Required Type Description
SAMPLE For multi-sample str Sample id; needed only when the table stacks several samples (select with sample_id=).
#CHR Yes str Chromosome.
START Yes int Start coordinate (0-indexed, inclusive).
END Yes int End coordinate (exclusive).
cn_(\w+) Yes (>= 1) str Per-clone allele copy number as "a|b", one column per clone (cn_normal, cn_clone1, ...).
u_(\w+) No float Per-clone proportion in [0, 1]; drives the proportion legend. Constant down the column.
PI_VIOL No bool Per-segment pure-integer-violation flag; drawn as a red/green overlay. Matched case-insensitively.

obs_df (Bin-level observations)

Bin-level observations. Consumed by plot_scatter_1d / plot_scatter_1d_multisample (ycol) and plot_scatter_2d (xcol, ycol).

Column Required Type Description
#CHR Yes str Chromosome.
START Yes int Start coordinate (0-indexed, inclusive).
END Yes int End coordinate (exclusive).
group For multi-sample str Group / sample id; column name set by group_col= (default SAMPLE).
value Yes (>= 1) float The plotted quantity, named freely and passed as ycol / xcol (e.g. RD, BAF, FCN-A). Used as given - no scaling or log.
hue No any Categorical column to color points by, named via hue= (e.g. cnp holding the joint "a|b;a|b" state).
keep_col No bool Marks bins that passed upstream QC; False bins are drawn in filtered_color. Column name set by keep_col=.

expected_df (1D overlay)

Expected value per segment (or bin) for the step overlay on plot_scatter_1d. May be coarser than obs_df - it maps through the same axis.

Column Required Type Description
#CHR Yes str Chromosome.
START Yes int Start coordinate (0-indexed, inclusive).
END Yes int End coordinate (exclusive).
exp_(\w+)_(\w+) Yes float Expected value, one exp_<value>_<group> column per value per group (e.g. exp_RD_S1). A missing column simply draws no overlay for that group.

expected_df (2D landmarks)

Expected copy-number landmarks for plot_scatter_2d; a seg.ucn-style table with precomputed landmark coordinates.

Column Required Type Description
SAMPLE For multi-sample str Group id; selected with group=.
#CHR, START, END No str/int Provenance only;
cn_(\w+) Yes str Per-clone "a|b" CN, one column per clone; sets each landmark's (a, b) label and dedup key.
u_(\w+) No float Per-clone proportion; drives the proportion legend.
exp_<xcol> Yes float Landmark x coordinate (matches the observed xcol).
exp_<ycol> Yes float Landmark y coordinate (matches the observed ycol).
is_filtered No bool Drop this landmark from the plot.
is_balanced No bool Draw a square marker instead of a circle.

Heatmap inputs

plot_heatmap / plot_heatmap_cnp take a reduced matrix, not a DataFrame, plus a bin table.

Input Required Type Description
matrix Yes ndarray (n_rows, n_bins) values; columns aligned to coords_df rows, rows bottom-to-top.
coords_df Yes DataFrame The bins: #CHR, START, END, one row per matrix column.
row_labels No ndarray (n_rows,) label per row; contiguous runs get a boxed y-tick.
strips No list[dict] Ordered annotation strips left of the heatmap, the first closest to it. Each dict is one of the two strip specs below.

Strip specs (entries of strips)

A categorical strip colors each row by its value; a distribution strip stacks each row's distribution (e.g. copy-typing posteriors) into a bar; a continuous strip colors each row by a scalar (e.g. purity). The kind is told apart by the data key: values marks categorical, matrix distribution, scalar continuous.

Key Kind Required Type Description
name all Yes str Strip key; also the default legend title and shared-palette identity.
display_name all No str Title shown on the strip and legend; defaults to name.
values categorical Yes ndarray (n_rows,) value per row.
cmap categorical No dict {value: color}; auto-built (shared across strips) when omitted.
matrix distribution Yes ndarray (n_rows, len(order)) per-row distribution, rows summing to 1.
order distribution Yes list Column order of matrix, i.e. the stacked categories.
cmap distribution Yes dict {category: color} for the stacked bars.
props distribution No dict {category: fraction} for the strip's legend.
scalar continuous Yes ndarray (n_rows,) scalar per row.
cmap continuous Yes Colormap Colormap (name or object) for the scalar.
norm continuous No Normalize Scalar normalization; matplotlib autoscales when omitted.
show_cbar continuous No bool Draw a colorbar for the scalar in the legend column (default off).
cbar_ticks continuous No list Explicit colorbar ticks when show_cbar is set.

Continuous strips carry no legend unless show_cbar is set (then a colorbar).