Lead overlap plot

Example

import gwaslab as gl
import pandas as pd

This tutorial builds small synthetic Sumstats objects and plots lead-locus overlap with gl.plot_lead_overlap(). For the full parameter reference, see Lead Overlap Plot.

Two-study Venn diagram

When comparing two studies, mode="auto" selects a Venn diagram. Leads within 500 kb on the same chromosome are merged into one locus group.

Example

def make_sumstats(rows, study_name):
    df = pd.DataFrame(rows)
    ss = gl.Sumstats(
        sumstats=df,
        snpid="SNPID",
        chrom="CHR",
        pos="POS",
        p="P",
        verbose=False,
    )
    ss.meta["gwaslab"]["study_name"] = study_name
    return ss

ss_a = make_sumstats(
    [
        {"SNPID": "s1_shared", "CHR": 1, "POS": 1_000_000, "P": 1e-10},
        {"SNPID": "s1_only", "CHR": 1, "POS": 10_000_000, "P": 1e-9},
    ],
    "Study A",
)
ss_b = make_sumstats(
    [
        {"SNPID": "s2_shared", "CHR": 1, "POS": 1_100_000, "P": 1e-12},
        {"SNPID": "s2_only", "CHR": 2, "POS": 2_000_000, "P": 1e-9},
    ],
    "Study B",
)

overlap_df, fig, log = gl.plot_lead_overlap(
    objects=[ss_a, ss_b],
    titles=["Study A", "Study B"],
    anno=False,
    windowsizekb_for_overlap=500,
    title="Lead locus overlap (2 studies)",
    verbose=False,
)
overlap_df[["LOCUS_ID", "MEMBERSHIP_KEY", "N_STUDIES", "STUDIES"]]

Two-study Venn diagram

Four-study UpSet plot

With four or more studies, mode="auto" switches to an UpSet matrix. Each row is a unique membership pattern (which studies share a locus).

Example

studies = []
for i in range(3):
    studies.append(
        make_sumstats(
            [{"SNPID": f"s{i}a", "CHR": 1, "POS": 1_000_000, "P": 1e-10}],
            f"GWAS{i + 1}",
        )
    )
studies.append(
    make_sumstats(
        [{"SNPID": "s4_only", "CHR": 3, "POS": 5_000_000, "P": 1e-9}],
        "GWAS4",
    )
)

overlap_df, fig, log = gl.plot_lead_overlap(
    objects=studies,
    titles=[f"GWAS{i + 1}" for i in range(4)],
    mode="auto",
    anno=False,
    windowsizekb_for_overlap=500,
    title="Lead locus overlap (4 studies, UpSet)",
    verbose=False,
)
set_list = overlap_df.attrs.get("set_list")
set_list

Four-study UpSet plot