Skip to content

Plot Gallery

Use plots after QC or fine-mapping, not as a replacement for checking the run summary. The fastest habit is: read the table, plot the suspicious locus, then decide whether to rerun.

Quick Commands

credtools plot work/qc/qc.txt.gz \
  --type summary \
  --output plots/qc_summary.png

credtools plot work/results/locus_1 \
  --type locusplot \
  --output plots/locus_1_finemap.png

credtools plot work/qc/locus_1 \
  --type locus_qc \
  --output plots/locus_1_qc.png

Auto-Detection

If you omit --type, CREDTOOLS tries to infer the plot:

Input Auto type
directory with pips.txt.gz locusplot
directory without pips.txt.gz locus_qc
file ending in qc.txt.gz or qc.txt summary
file ending in compare_maf.txt.gz maf_corr

When auto-detection fails, pass --type explicitly.

Available Plot Types

  • summary

    Input: qc.txt.gz

    Shows lambda-s, MAF correlation, lambda-s outlier counts, and Dentist-S outlier counts in one figure.

  • locus_qc

    Input: a per-locus QC directory

    Combines locus-level QC panels. Use this for the worst loci from the summary table.

  • locusplot

    Input: a fine-mapping locus directory with pips.txt.gz

    Shows p-values, LD-to-lead coloring, PIP-scaled points, and credible-set membership.

  • lambda_s

    Input: qc.txt.gz

    Boxplot of lambda_s by cohort.

  • maf_corr

    Input: qc.txt.gz or compare_maf.txt.gz

    Barplot of MAF correlation by cohort.

  • lambda_s_outliers

    Input: qc.txt.gz

    Count of kriging RSS outliers by cohort.

  • dentist_s_outliers

    Input: qc.txt.gz

    Count of Dentist-S outliers by cohort.

  • locus_pvalues

    Input: expected_z.txt.gz

    P-value style view from z-scores, optionally with credible-set annotations when called through Python.

  • zscore_qq

    Input: expected_z.txt.gz

    QQ plot for observed z-score behavior.

  • ld_decay

    Input: ld_decay.txt.gz

    LD decay by distance.

  • ld_4th_moment

    Input: ld_4th_moment.txt.gz

    Per-variant LD fourth-moment comparison.

  • snp_missingness

    Input: snp_missingness.txt.gz

    UpSet-style view of variant presence across cohorts.

Size and Format

Use --width, --height, and --dpi for reports:

credtools plot work/qc/qc.txt.gz \
  --type summary \
  --width 14 \
  --height 10 \
  --dpi 300 \
  --output plots/qc_summary.pdf

The output extension controls the format. Common choices are .png, .pdf, and .svg.

Locus Plot Requirements

locusplot reads pips.txt.gz and expects:

Column pattern Why it matters
BP x-axis position
PIP point size
CRED credible-set outline
columns ending with _P cohort-specific p-values
matching columns ending with _R2 LD color scale

If a locus plot fails with a missing _R2 column, the PIP table was not created from a fully matched summary-statistics and LD input.

Suggested Review Set

For a large run, do not plot every locus at first. Plot:

  • the locus with the highest lambda_s,
  • the locus with the lowest maf_corr,
  • the locus with the most Dentist-S outliers,
  • two or three high-priority biological loci,
  • one clean-looking locus as a baseline.