perf: "two-pass" seurat hvg via scanpy.get.aggregate

[ilan-gold]

An idea that popped into my head for disk-bound datasets but likely also normal ones. This should, in theory, greatly improve on-disk access and produce speed ups for disk bound data by reducing the amount of i/o in the worst case, unordered scenario (while, I would guess, leaving in-memory datasets untocuhed or maybe improved thanks to memory access + more efficient mean/var).

• Closes #
• Tests included or not required because:

• Release notes not necessary because:

[codecov]

❌ 1 Tests Failed:

Tests completed	Failed	Passed	Skipped
1638	1	1637	1103

View the top 1 failed test(s) by shortest run time

tests/test_neighbors.py::test_connectivities_euclidean[umap]

Stack Traces | 0.009s run time

[XPASS(strict)] umap<0.6 is broken with numba≥0.62.0rc1

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}

[scverse-benchmark]

Benchmark changes

Change	Before [9bc2c1e]	After [7e0390e]	Ratio	Benchmark (Parameter)
+	1.09±0s	2.39±0s	2.19	preprocessing_log.HVGSuite.time_highly_variable_genes('seurat_v3')

Comparison: https://github.com/scverse/scanpy/compare/9bc2c1ed1bd1cf6f6c06cec71cce99916d048163..7e0390ee10fe2c38d382b97ad2ff0bf8e6280e6b
Last changed: Thu, 9 Apr 2026 08:15:10 +0000

More details: https://github.com/scverse/scanpy/pull/4013/checks?check_run_id=70564424668