Skip to content

Variant set plot

Example

import gwaslab as gl
import pandas as pd

This tutorial shows how to compare effect sizes for a fixed set of variants across multiple GWAS studies using gl.SumstatsSet and plot_effect(). For the full parameter reference, see Variant Set Plot.

In development

SumstatsSet and plot_effect() are under active development. The development notebook lives at examples/_development/variant_set/visualization_variant_set.ipynb.

Basic variant set plot

Load two or more Sumstats objects, pass a list of variant IDs to SumstatsSet, then call plot_effect() to draw a forest-style panel grouped by variant.

Example

rows_a = [
    {"CHR": 1, "POS": 100, "EA": "A", "NEA": "G", "BETA": 0.10, "SE": 0.02, "P": 1e-6, "EAF": 0.35, "SNPID": "1:100:A:G"},
    {"CHR": 2, "POS": 200, "EA": "C", "NEA": "T", "BETA": -0.05, "SE": 0.03, "P": 1e-4, "EAF": 0.22, "SNPID": "2:200:C:T"},
    {"CHR": 7, "POS": 500000, "EA": "G", "NEA": "A", "BETA": 0.15, "SE": 0.025, "P": 1e-8, "EAF": 0.18, "SNPID": "rs7890"},
]
rows_b = [
    {"CHR": 1, "POS": 100, "EA": "A", "NEA": "G", "BETA": 0.08, "SE": 0.02, "P": 5e-8, "EAF": 0.33, "SNPID": "1:100:A:G"},
    {"CHR": 7, "POS": 500000, "EA": "G", "NEA": "A", "BETA": 0.12, "SE": 0.03, "P": 1e-6, "EAF": 0.20, "SNPID": "rs7890"},
]
rows_c = [
    {"CHR": 1, "POS": 100, "EA": "A", "NEA": "G", "BETA": 0.06, "SE": 0.025, "P": 1e-4, "EAF": 0.30, "SNPID": "1:100:A:G"},
    {"CHR": 7, "POS": 500000, "EA": "G", "NEA": "A", "BETA": 0.09, "SE": 0.028, "P": 1e-5, "EAF": 0.16, "SNPID": "rs7890"},
]

def make_sumstats(rows):
    return gl.Sumstats(
        sumstats=pd.DataFrame(rows),
        chrom="CHR", pos="POS", ea="EA", nea="NEA",
        snpid="SNPID", beta="BETA", se="SE", p="P", eaf="EAF",
        verbose=False,
    )

studies = {"StudyA": make_sumstats(rows_a), "StudyB": make_sumstats(rows_b), "StudyC": make_sumstats(rows_c)}
variant_set = [[1, 100], "2:200:C:T", "rs7890"]

sset = gl.SumstatsSet(studies, variant_set=variant_set, verbose=False)
sset.data[["STUDY", "SNPID", "BETA", "SE", "EAF"]]

Example

sset.plot_effect(
    y="STUDY",
    group=["CHR", "POS"],
    y_sort=["CHR", "POS", "STUDY"],
    hue="STUDY",
    gap=1,
    title="Variant set: effect sizes across studies",
    effect_label="Per-allele effect size",
    save="variant_set_basic.png",
    verbose=False,
)

Basic variant set plot

Variant ID formats

variant_set accepts several identifier styles. Matching uses SNPID for strings and CHR + POS for coordinate pairs.

Format Example Matches by
rsID "rs7890" Exact SNPID
CHR + POS [1, 100] CHR and POS
Variant ID string "1:100:A:G" Parsed CHR and POS

Example

variant_set = ["rs7890", [1, 100], "2:200:C:T"]
sset = gl.SumstatsSet(studies, variant_set=variant_set, verbose=False)

Load studies from a glob pattern

When many cohort files share the same column layout, pass a glob pattern instead of a dictionary. Study names are derived from filenames (e.g. cohort_EUR.txt → cohort_EUR).

Example

sset = gl.SumstatsSet(
    "./data/cohort_*.txt",
    variant_set=["rs7890", [1, 100]],
    fmt="auto",
    build="19",
    verbose=False,
)
sset.plot_effect(
    y="STUDY",
    group=["CHR", "POS"],
    y_sort=["CHR", "POS", "STUDY"],
    hue="STUDY",
    eaf_panel=True,
    save="variant_set_glob.png",
    verbose=False,
)

Glob-loaded variant set plot

Side panels: EAF and SNPR2

When EAF or SNPR2 columns are present, plot_effect() adds optional horizontal bar panels beside the main effect panel. Panels are skipped automatically if the column is missing.

Panel Column Description
EAF EAF Effect-allele frequency
SNPR2 SNPR2 Per-variant R² (see Per-SNP R2 and F)

Disable a panel with eaf_panel=False or snpvar_panel=False.

Function Use case
compare_effect() Scatter comparing two studies (auto lead extraction)
plot_forest() Meta-analysis forest plot with combined estimate
plot_lead_overlap() Venn/UpSet of lead locus overlap across studies