Quickstart

Inspect a dataset

import pandas as pd
from mismap_qc import qc

df = pd.read_csv("proteomics.tsv", sep="\t", index_col=0)

report = qc(df, group_level="condition")
print(report)
# MismapReport(n=8412x96, 3 outliers, 412 MNAR features, passed=True)

report is a frozen snapshot. Re-run qc() to get a new one; it never mutates.

For a readable multi-section view rather than the one-line repr:

print(report.summary())

Drill into what was flagged

The report holds pandas DataFrames, so anything flagged can be queried directly:

report.sample_outliers.query("flagged")
report.feature_mechanism.query("mechanism == 'MNAR'")

Gate a pipeline

assert_qc() raises MismapQCFailure when a rule is violated:

from mismap_qc import assert_qc

assert_qc(df, thresholds={
    "min_sample_completeness": 0.60,
    "max_mnar_fraction": 0.30,
    "max_sample_outliers": 3,
})

Rules carry a default severity of error, warning, or info. Warning-severity violations emit MismapQCWarning through the standard warnings module rather than raising, and can be silenced the usual way:

import warnings
from mismap_qc import MismapQCWarning

warnings.filterwarnings("ignore", category=MismapQCWarning)

Override a rule's severity per call with severity_overrides.

Get the numbers behind any plot

Every plot function accepts return_data=True and returns (Figure, DataFrame) against a documented schema:

from mismap_qc import detection_waterfall

fig, table = detection_waterfall(df, return_data=True)
# table columns: feature, detection_rate, rank

This matters when a plot truncates. missing_upset() caps the figure at the 50 largest intersections, but the returned table contains every intersection with a plotted column recording which ones made the figure, so nothing is hidden:

from mismap_qc import missing_upset

fig, table = missing_upset(df, return_data=True)
table.query("~plotted")                       # what the figure left out
table.query("members == 'S1|S3'").feature     # features missing in exactly S1 and S3

Start from AnnData

from mismap_qc import from_anndata, qc

df = from_anndata(adata, obs_levels=["batch", "condition"])
report = qc(df, group_level="condition")

from_anndata() needs the optional anndata extra: pip install mismap-qc[anndata].