Sync wiki: section crops include the domain header
Sync wiki: link_template is automatic from the ID (fix stale docs)
Sync wiki: 3-route flow diagram (plot / genome browser / analysis)
Host images in the wiki repo + relative paths: fixes intermittent/broken images (was proxied from raw.githubusercontent via camo, rate-limited)
Sync wiki: per-section output crops in Mapping; auto linkout + Static/Interactive table in Plotting; public viewer source in Python-API; two-route flow diagram
Sync wiki: clean node-based flow diagram
Sync wiki: harmonized flow diagram
Sync wiki: larger gallery images + static multi-isoform stack
Sync wiki: drop maintainer-only gallery-regeneration block
Sync wiki from main: single --out/--engine plotting, viewer GIFs, ASCII prose, updated benchmarks/figures
Notebooks: consolidate atlas+ClinVar, redesigned analyses; perf threads {1,2,4,8,16}; drop removed-notebook links
Index: add mouse-vm25 pre-built index
Perf: one headline (~970x end-to-end), drop warm-map multipliers; short Docker section
Architecture: drop removed test_notebooks.py
Architecture: drop reference to the local-only notebook generator
Repo reorg: parsing/, software_tests/, tutorial/{examples,reproduce_paper/{benchmarks,end_to_end}} — update all paths
Mapping: trim column-4 note, one unified Feature-TSV column table; move query-BED-building to FAQ
fetch Zenodo-only wording; plain Windows note; drop Docker/smoke; consolidate benchmark tables (1M threads×batch grid); custom-GTF stdout-only
Updated Mapping (markdown)
Updated Installation (markdown)
Updated Home (markdown)
Fixed Typos
Performance page: VisProDom/geneplot numbers now on the human v86 set Both run on the same human Ensembl-86 set instead of their bundled maize/fruit-fly examples: VisProDom 42.7 s / 2.3 GB (rebuilds the genome per call), geneplot ~14 genes/s (gffutils DB + per-gene domain re-read).
Performance page: fresh GenomicFeatures/VisProDom numbers, correct geneplot (it does map)
Rename benchmarking.ipynb link to scaling_and_ram.ipynb (self-benchmark)
Point Performance page at validation/software_comparison notebooks; add walkthrough nbviewer link The combined proteintogenome_comparison.ipynb is being split into validation.ipynb (accuracy) + software_comparison.ipynb (speed); repoint the accuracy and speed sections accordingly. Add a walkthrough nbviewer link on the Tutorials page since its interactive viewers don't render on GitHub.
Fix index docs: Home fetch path, drop bogus human-v115 row, 298 MB index, Plotting segments import - Home quickstart: 'fetch human --out human.idx' so the next map step's --index human.idx actually resolves (fetch defaults to ~/.cache). - Drop the fabricated 'human-v115 / ensembl_v15_human.idx' pre-built row (no such Zenodo target). - Performance: human GENCODE index is ~298 MB, not 128 MB. - Plotting: import _segments_from_dataframe from prot2exon.plot instead of the broken p2e.plot.<attr> path.
Point testing docs at pytest, not the removed run_tests.py The monolithic tests/run_tests.py was migrated into the pytest suite, so update Installation/Architecture/FAQ to describe pytest (modules, golden files, --update-goldens) and the pip install -e + pytest -q run.
Clarify fetch as Zenodo-download-only in the index docs fetch read as having two meanings (download a pre-built .idx vs. wget + build an arbitrary GTF), which confused readers. Reframe section 3 as a pure Zenodo download: lead with --out so the index lands where you run, not hidden in the cache, and route non-Zenodo releases back to the manual obtain-GTF -> prot2exon index path (with the build shortcut noted once, pointing at --help).
Add "The same in Python:" lead-in before mirrored Python blocks In Mapping.md the per-output-mode Python snippets followed the CLI command with no separating text, reading as if both had to be run. Add a short lead-in so each Python block clearly shows the equivalent, not an additional step.