550a0e2666bcce27181a0cb0eab32ed261e444f6 mspeir Thu Aug 27 14:56:42 2026 -0700 singleCellSignalsPeaks: color the Cell class facet checkboxes, refs #37820, refs #37914 The faceted UI already draws a color swatch beside each checkbox of any facet named in a colorSettingsUrl JSON, and the track already colors its subtracks by broad cell class from celltype-palette.tsv. Publish that palette so the selector shows the same colors: copySingleCellSignalsPeaksFiles.py now writes <bed>/singleCellSignalsPeaks_colors.json alongside the facet metadata, and the composite header names it. Rendered from the one shared palette, so a class is the same color in the checkbox list, in the drawn tracks, and on both assemblies. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> diff --git src/hg/makeDb/scripts/singleCellSignalsPeaks/copySingleCellSignalsPeaksFiles.py src/hg/makeDb/scripts/singleCellSignalsPeaks/copySingleCellSignalsPeaksFiles.py index 289f4a40d85..cbf63d44b87 100644 --- src/hg/makeDb/scripts/singleCellSignalsPeaks/copySingleCellSignalsPeaksFiles.py +++ src/hg/makeDb/scripts/singleCellSignalsPeaks/copySingleCellSignalsPeaksFiles.py @@ -1,77 +1,132 @@ #!/usr/bin/env python3 """ Copy the source data files for the native singleCellSignalsPeaks track into the genome's bed directory, mirroring each file's served relative path (Redmine #37820 for hg38, #37914 for mm10). For the given assembly it reads the hub build's main signal-&-peaks composite (cellBrowser<Asm>) stanzas, and for every subtrack copies the source file (resolved from the hub manifest by served relative path) to /hive/data/genomes/<asm>/bed/singleCellSignalsPeaks/<served-relpath> The served subpath is preserved on purpose: some peak-file basenames repeat across datasets, so a flat directory would clobber them, and keeping the subpath lets the /gbdb/<asm>/bbi/singleCellSignalsPeaks symlink resolve every bigDataUrl. It also copies the composite's facet metadata to - <bed>/singleCellSignalsPeaks_metadata.tsv (the track's metaDataUrl target). + <bed>/singleCellSignalsPeaks_metadata.tsv (the track's metaDataUrl target), +and writes the cell-class facet colors to + <bed>/singleCellSignalsPeaks_colors.json (the track's colorSettingsUrl target). Usage: copySingleCellSignalsPeaksFiles.py [--assembly hg38|mm10] [--stanzas STANZAS] [--manifest MANIFEST] [--dry-run] """ -import re, os, shutil, argparse +import re, os, json, shutil, argparse from urllib.parse import urlparse # Where the hub build writes manifest.tsv -- its OUTPUT dir, not its code. The build # itself lives in the cellBrowser repo (ucsc/allTracksHub): # https://github.com/ucscGenomeBrowser/cellBrowser/tree/develop/ucsc/allTracksHub # Its output dir is set there by CBHUB_OUT; keep this default in step with it (or pass # --manifest). HUB_BUILD = os.environ.get( "HUB_BUILD", "/hive/data/inside/cells/all-tracks-hub-build") TRACK = "singleCellSignalsPeaks" def copy_atomic(src, dst): """Copy src to dst without ever leaving a half-written dst in place. The bed dir is served live -- /gbdb/<asm>/bbi/singleCellSignalsPeaks is a symlink straight to it -- so copying onto a file the browser may be reading hands out a truncated bigBed for as long as the copy takes. Write a temp file beside the destination and rename it, which is atomic within one filesystem. """ tmp = "%s.tmp%d" % (dst, os.getpid()) try: shutil.copy2(src, tmp) os.replace(tmp, dst) except BaseException: if os.path.exists(tmp): os.remove(tmp) raise +def write_text_atomic(text, dst): + """Write text to dst without ever leaving a half-written dst in place. + + Same reason as copy_atomic: the bed dir is served live, and the faceted UI fetches + this file on every hgTrackUi page load, so a partial write hands out unparseable JSON + for as long as the write takes. + """ + tmp = "%s.tmp%d" % (dst, os.getpid()) + try: + with open(tmp, "w") as fh: + fh.write(text) + os.replace(tmp, dst) + except BaseException: + if os.path.exists(tmp): + os.remove(tmp) + raise + def up_to_date(src, dst): """True if dst already holds this copy of src, so it can be skipped. Size alone is not enough: a rebuilt file often lands on the same size, and skipping it then serves last month's data forever. shutil.copy2 carries the mtime across, so a dst older than its source means the source moved on. """ if not os.path.exists(dst): return False return (os.path.getsize(dst) == os.path.getsize(src) and os.path.getmtime(dst) >= os.path.getmtime(src)) +def palette_file(build): + """Path to the cell class -> color palette TSV (class<TAB>R,G,B, one per line). + + Same resolution order as makeSingleCellSignalsPeaksRa.py: prefer the copy of record + archived alongside these scripts (written by build_celltype_crosswalks.py), fall back + to the hub build dir. The two scripts must agree on the palette, or the swatches in + the faceted selector would not match the colors the subtracks are drawn in. + """ + pal = os.path.join(os.path.dirname(os.path.abspath(__file__)), + "celltype-crosswalks", "celltype-palette.tsv") + if not os.path.isfile(pal): + pal = os.path.join(build, "celltype-crosswalks", "celltype-palette.tsv") + return pal + +def colors_json(pal): + """Render the palette as the faceted UI's colorSettingsUrl JSON. + + facetedComposite.js wants {facetName: {facetValue: cssColor}} and draws a swatch + beside each checkbox of any facet named in it. The facet name must be the metadata + column ("Cell_class") and each key must be the column value verbatim -- the lookup + is an exact string match, so a case or spacing difference silently drops the swatch. + The whole palette is emitted, not just the classes this assembly uses: extra keys are + ignored by the JS, and it keeps hg38 and mm10 sharing one class->color mapping. + """ + colors = {} + for line in open(pal): + f = line.rstrip("\n").split("\t") + if len(f) < 2 or not f[0].strip(): + continue + rgb = [int(x) for x in f[1].split(",")] + colors[f[0]] = "#%02X%02X%02X" % tuple(rgb) + if not colors: + raise SystemExit("ERROR: no class/color rows read from %s, so the facet color " + "swatches cannot be written." % pal) + return json.dumps({"Cell_class": colors}, indent=4, sort_keys=True) + "\n" + def load_relpath_to_abs(manifest, asm): m = {} with open(manifest) as fh: hdr = fh.readline().rstrip("\n").split("\t") ai, ui, asmi = hdr.index("abs_path"), hdr.index("track_url"), hdr.index("assembly") for line in fh: f = line.rstrip("\n").split("\t") if len(f) <= max(ai, ui, asmi) or f[asmi] != asm: continue m[urlparse(f[ui]).path.lstrip("/")] = f[ai] return m def main(): ap = argparse.ArgumentParser() ap.add_argument("--assembly", default="hg38", choices=["hg38", "mm10"]) @@ -145,25 +200,40 @@ "manifest come from the same build?" % (total, hub_composite, stanzas, manifest)) # Missing metadata is fatal, the same way it is in makeSingleCellSignalsPeaksRa.py. # Skipping it quietly leaves the bed dir advertising the previous build's facets # against this build's data files, which is the mismatch nobody would go looking for. meta_src = os.path.join(build, "meta", "%s.metadata.tsv" % asm) meta_dst = os.path.join(beddir, "%s_metadata.tsv" % TRACK) if not os.path.isfile(meta_src): raise SystemExit("ERROR: no facet metadata at %s, so %s cannot be refreshed. The " "track's metaDataUrl would keep pointing at the previous build's " "facets." % (meta_src, meta_dst)) if not args.dry_run: os.makedirs(beddir, exist_ok=True) copy_atomic(meta_src, meta_dst) + + # Facet color swatches: the track's colorSettingsUrl target. Derived from the same + # class->color palette the subtrack "color" lines come from, so a class has one color + # in the selector and in the track display. A missing palette is fatal for the same + # reason a missing metadata file is: the .ra names this file, and quietly leaving the + # previous build's copy in place would show swatches that no longer match the tracks. + pal = palette_file(build) + if not os.path.isfile(pal): + raise SystemExit("ERROR: no cell class palette at %s, so the facet color swatches " + "cannot be written. The track's colorSettingsUrl would keep " + "pointing at the previous build's colors." % pal) + colors_dst = os.path.join(beddir, "%s_colors.json" % TRACK) + if not args.dry_run: + write_text_atomic(colors_json(pal), colors_dst) + print("assembly=%s composite=%s: subtracks=%d copied=%d missing=%d ~%.1f GB%s" % (asm, hub_composite, total, copied, missing, nbytes / 1e9, " (dry-run)" if args.dry_run else " -> " + beddir)) if misses: print("MISSING %d source files:" % len(misses)) for r in misses[:25]: print(" " + r) if __name__ == "__main__": main()