I tiakina i:
Ngā taipitopito rārangi puna kōrero
Ngā kaituhi matua: AdamNogell, Eda Ehler, Ondřej Moravčík, satrovam
Hōputu: Recurso digital
Reo:
I whakaputaina: Zenodo 2026
Urunga tuihono:https://doi.org/10.5281/zenodo.20284071
Ngā Tūtohu: Tāpirihia he Tūtohu
Kāore He Tūtohu, Me noho koe te mea tuatahi ki te tūtohu i tēnei pūkete!
Rārangi ihirangi:
  • <p>Update containing small bug fixes and triggering Zenodo archiving. No breaking API changes.</p> <h3>Fixes</h3> <ul> <li><p><strong><code>extract_top_regions</code> overlap & boundary handling</strong> (<code>plot.py</code>)</p> <ul> <li><code>min_gap</code> is now an <strong>edge-to-edge</strong> gap between window intervals, not a center-to-center distance. Adjacent peaks no longer produce overlapping or touching regions.</li> <li><code>start</code> and <code>end</code> are now the true <code>±window_size/2</code> boundaries around <code>center</code>, so they remain well-defined even when only a single column position falls inside the window (e.g. pre-summarised 25 kb Roadmap windows).</li> <li>Stats are NaN-safe: windows that are entirely NaN are skipped, and <code>mean_score</code> / <code>max_score</code> use <code>nanmean</code> / <code>nanmax</code>.</li> <li>New column <code>n_positions</code>; <code>n_variants</code> is kept as a legacy alias for backwards compatibility.</li> </ul> </li> <li><p><strong>Interactive heatmap downsampling</strong> (<code>interactive.py</code>)</p> <ul> <li>Replaced integer-division binning with <code>np.array_split</code>, so all columns are distributed evenly across bins and trailing positions are no longer dropped or stuffed into the last bin when <code>n_cols</code> isn't a multiple of <code>max_columns</code>.</li> </ul> </li> <li><p><strong>Colorbar kwargs normalization</strong> (<code>plot.py</code>)</p> <ul> <li><code>cbar_kws</code> no longer gets an empty <code>ticks</code> entry attached when <code>cbar_ticks</code> is unset, making behavior consistent across input types.</li> </ul> </li> <li><p><strong>Error handling</strong></p> <ul> <li><code>plot.py</code>: replaced bare <code>except:</code> around <code>start:end</code> cropping with a targeted <code>except KeyError:</code> — unexpected errors are no longer silently swallowed.</li> <li><code>prepare.py</code>: VCF→Zarr conversion errors now log with full traceback (<code>exc_info=True</code>) for easier debugging.</li> <li><code>rank_tools.py</code>: <code>compute_ranks</code> now <strong>raises</strong> <code>ValueError</code> with the mismatched counts instead of only logging and returning <code>None</code>.</li> </ul> </li> </ul>