metafopy

Meta-analysis in Python: a re-implementation of the R package metafor (reference version 5.2-1).

Function and argument names follow metafor, with . replaced by _ (rma.uni becomes rma_uni).

Numerical results are checked against metafor fixtures generated from R.

from metafopy import datasets, escalc, rma

dat = escalc(measure='RR', ai='tpos', bi='tneg', ci='cpos', di='cneg', data=datasets.load('dat.bcg'))
print(rma('yi', 'vi', data=dat))

Random-Effects Model (k = 13; tau^2 estimator: REML)

tau^2 (estimated amount of total heterogeneity): 0.3132 (SE = 0.1664)
tau (square root of estimated tau^2 value):      0.5597
I^2 (total heterogeneity / total variability):   92.22%
H^2 (total variability / sampling variability):  12.86

Test for Heterogeneity:
Q(df = 12) = 152.2330, p-val < .0001

Model Results:

estimate      se     zval    pval    ci.lb    ci.ub
 -0.7145  0.1798  -3.9744  <.0001  -1.0669  -0.3622  ***

---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Where to start

• Installation for requirements and a check that the install works.
• Getting started for a complete first analysis, from raw 2x2 tables to a forest plot.
• The user guide, for one task at a time:
• Effect sizes
• Fitting models
• Binary and count data
• Multilevel and multivariate models
• Selection models and location-scale models
• Diagnostics
• Publication bias
• Plots
• Plot gallery, the plots of the metafor documentation next to those of metafopy
• Datasets
• Analyses for worked examples that mirror well-known metafor analyses.
• Coming from R if you already know metafor.
• The function reference, generated from the docstrings.
• Differences from metafor for the places where metafopy and metafor do not agree.

Why metafopy

metafor is the most complete meta-analysis package available, but it lives in R. metafopy gives Python users the same models with the same names, defaults and printed output, so the metafor documentation, the metafor analyses gallery and published code translate directly. Every fitted model is a plain, frozen dataclass of NumPy arrays, so results drop into the rest of the scientific Python stack.

The trade-off is that metafopy follows metafor, including its conventions and its defaults (REML for the heterogeneity estimate, Wald-type tests, effect sizes on the log scale for ratios). Where Python or NumPy force a different design, the differences page says so.

License

GPL-2.0-or-later, the same as metafor.

The bundled datasets come from the R package metadat.