[padix-key]

GROFile.get_structure() parsed each atom line in a Python loop and assigned every value individually into NumPy arrays (7 scalar __setitem__ calls per atom). In addition, _is_int() (used by get_structure() and get_model_count() to find model starts) raised and caught a ValueError for every atom line.

Fix:

• Parse annotation and coordinate columns with list comprehensions and assign each column at once.
• _is_int() returns False early for lines containing a ., which int() can never parse (all atom and box lines), avoiding exception handling.

Results are identical (values and dtypes checked against main).

Input	Function	before	after	speedup
1aki.gro (1k atoms)	get_structure(model=1)	3.23 ms	1.18 ms	2.7x
1l2y.gro (38 models x 304 atoms)	get_structure()	0.99 ms	0.41 ms	2.4x
1igy.gro	get_structure(model=1)	37.5 ms	14.0 ms	2.7x
waterbox.gro	get_structure(model=1)	11.4 ms	2.4 ms	4.7x
1aki.gro	get_model_count()	0.67 ms	0.046 ms	14.6x

Note: for malformed multi-model files where a later model has a different atom count, a ValueError (broadcasting) is now raised instead of an IndexError/silently partially filled coordinates.

Tests run: tests/structure/io/test_gro.py.

Authored by Claude

[codspeed]

Merging this PR will regress 3 benchmarks

⚡ 14 improved benchmarks
❌ 3 regressed benchmarks
✅ 92 untouched benchmarks
🆕 2 new benchmarks
⏩ 14 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	benchmark_match_kmer_selection[BucketKmerTable(10000)-11*11*1*1***111]	648.6 µs	771.5 µs	-15.93%
❌	benchmark_cell_list_compute_contacts[MAPPING]	22.9 ms	26.7 ms	-14.32%
❌	benchmark_cell_list_compute_contacts[PAIRS]	22.7 ms	25.8 ms	-12.11%
⚡	benchmark_find_rotatable_bonds	177.1 ms	26.7 ms	×6.6
⚡	benchmark_superimpose[superimpose]	3,198.9 µs	647.7 µs	×4.9
⚡	benchmark_hbond	29.7 ms	10.1 ms	×2.9
⚡	benchmark_similar_kmers	483 µs	201.8 µs	×2.4
⚡	benchmark_superimpose_multi_model[superimpose]	14.4 ms	6.4 ms	×2.3
⚡	benchmark_get_assembly	1,594.7 ms	836 ms	+90.77%
⚡	benchmark_superimpose[superimpose_homologs]	60.6 ms	33.6 ms	+80.49%
⚡	benchmark_superimpose_multi_model[superimpose_homologs]	74.2 ms	44.8 ms	+65.69%
⚡	benchmark_score_threshold_rule[None]	3.3 s	2 s	+65.28%
⚡	benchmark_score_threshold_rule[11*11*1*1***111]	3.3 s	2 s	+65.11%
⚡	benchmark_superimpose[superimpose_without_outliers]	31.9 ms	20.8 ms	+53.31%
⚡	benchmark_superimpose[superimpose_structural_homologs]	53.4 ms	44.8 ms	+19.08%
⚡	benchmark_set_structure[bcif-True]	39 ms	32.8 ms	+18.95%
⚡	benchmark_set_structure[bcif-False]	17.7 ms	15.1 ms	+16.74%
🆕	benchmark_stack[10]	N/A	6.7 ms	N/A
🆕	benchmark_stack[2]	N/A	2 ms	N/A

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing padix-key:perf/strucio-gro-read (685fbd5) with main (81df912)²}

Footnotes

1. 14 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

2. No successful run was found on main (b0d32ca) during the generation of this report, so 81df912 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩