49de9e93e4417083feb23522a71e3960595828a3 markd Fri Sep 25 22:46:28 2026 -0700 ProCapNet label and facet color fixes from QA. refs #35528 The composite longLabel named the sequence-contribution scores, which only exist on hg38, so it was wrong on hs1. Use "ProCapNet predicted PRO-cap" on both assemblies; the scores are described on the track description page. The Sample class swatches reused two colors from the cell-line palette, so a track's color could be read as its class: A673 is a cancer line and drew in #0072B2, which was the Non-cancer swatch. Sample class is a binary facet and does not need a hue, so use black and gray, outside the Okabe-Ito palette. Say in the description that the contribution scores cover about 1% of the genome, so the limit is visible before the display conventions section. diff --git src/hg/makeDb/outside/proCapNet/proCapNetTrackDb src/hg/makeDb/outside/proCapNet/proCapNetTrackDb index 4671744e4e8..01254905c4c 100755 --- src/hg/makeDb/outside/proCapNet/proCapNetTrackDb +++ src/hg/makeDb/outside/proCapNet/proCapNetTrackDb @@ -1,245 +1,248 @@ #!/usr/bin/env python3 """Generate the transcriptionStart.ra trackDb file and the faceted-composite metadata tables for one assembly. Layout is a superTrack holding one faceted composite per data source: proCapNet for the model predictions and sequence-contribution scores, encode4ProCap for the experimental PRO-cap signal. hs1 has predictions only. The composites must be faceted rather than plain. A container multiWig may be a composite child, but hui.c compositeUiSubtracks() walks descendant leaves, so under a plain composite the multiWig is flattened away and never drawn. A faceted composite is routed to facetedCompositeUi() instead, which builds its rows from metaDataUrl and leaves the multiWig intact. fiberSeqCompendium is the same pattern. Subtrack names are <composite>_<primaryKey>_<dataType>, which is how the faceted UI maps a table cell to a track. """ import argparse import json from pycbio.sys import cli, fileOps from pycbio.tsv import TsvReader GBDB = "/gbdb/{db}/{track}" DOWNLOAD = "https://hgdownload.soe.ucsc.edu/gbdb/{db}/{track}/$$" SAMPLE_CLASS_DESC = "Cancer = cell line derived from a tumor" -SAMPLE_CLASS_COLORS = {"Cancer": "#D55E00", "Non-cancer": "#0072B2"} +# Deliberately outside the Okabe-Ito palette the cell lines use. Sharing a hue +# with a cell line makes a track's color read as its sample class: A673 is a +# cancer line and draws in #0072B2, which was the Non-cancer swatch. +SAMPLE_CLASS_COLORS = {"Cancer": "#000000", "Non-cancer": "#808080"} def parseArgs(): parser = argparse.ArgumentParser(description=__doc__) parser.add_argument("db") parser.add_argument("experimentsTsv") parser.add_argument("outRa") parser.add_argument("gbdbDir", help="/gbdb directory holding the track directories, where " "the metadata and color files are written") return cli.parseOptsArgsWithLogging(parser) def stanza(indent, lines): pad = " " * indent return "".join(pad + line + "\n" for line in lines) + "\n" def downloadCell(db, track, exp): """The Files column, one comma-separated token per bigWig. Each token is path|label: the path substitutes into the subtrackUrls $$ and the label is what the cell shows, so the table offers a named link per file rather than a bare file name.""" if track == "proCapNet": files = [(f"pred/{exp.cell}.proCapNet.pos.bw", "+ strand"), (f"pred/{exp.cell}.proCapNet.neg.bw", "- strand")] if db == "hg38": files.append((f"contrib/{exp.cell}.proCapNet-contrib.bw", "contribution")) else: files = [(f"{exp.cell}.{exp.procapAcc}.pos.bw", "+ strand"), (f"{exp.cell}.{exp.procapAcc}.neg.bw", "- strand")] return ",".join(f"{path}|{label}" for path, label in files) def writeMetadata(path, experiments, accField, db, track): """The cell line table shown on the faceted track UI page. A plain column name gets facet checkboxes, a leading underscore means searchable and sortable but not faceted. facetedComposite.js only offers a facet value that occurs more than once, so Sample_class is the only column worth faceting: cell, tissue and accession are unique per row and would each draw six checkboxes that match one row apiece. Cell_line is the primaryKey, so it names the tracks and is shown as its own column; there is no second column repeating the cell name. Column names are underscore separated because toTitleStyle() in facetedComposite.js turns an underscore into a space but does not split camelCase. A header cell may carry a description after a "|", shown behind an info icon.""" fileOps.ensureFileDir(path) with fileOps.AtomicFileOpen(path) as fh: print(f"_Tissue\tSample_class|{SAMPLE_CLASS_DESC}\tExperiment\tFiles\tCell_line", file=fh) for exp in experiments: print(f"{exp.tissue}\t{exp.sampleClass}\t{getattr(exp, accField)}\t" f"{downloadCell(db, track, exp)}\t{exp.cell}", file=fh) def writeColors(path): "swatches beside the facet checkboxes, keyed by the faceted column name" fileOps.ensureFileDir(path) with fileOps.AtomicFileOpen(path) as fh: json.dump({"Sample_class": SAMPLE_CLASS_COLORS}, fh, indent=4) print(file=fh) def superStanza(): return stanza(0, [ "track transcriptionStart", "superTrack on show", "shortLabel Transcription Initiation (TSS)", "longLabel Transcription Initiation (TSS)", "group rna", ]) def compositeStanza(db, track, shortLabel, longLabel, dataTypes, accUrl, priority): gbdb = GBDB.format(db=db, track=track) return stanza(4, [ f"track {track}", "parent transcriptionStart", "compositeTrack faceted", "type bigWig", f"shortLabel {shortLabel}", f"longLabel {longLabel}", f"metaDataUrl {gbdb}/metadata.tsv", f"colorSettingsUrl {gbdb}/colors.json", "primaryKey Cell_line", f"dataTypes {dataTypes}", f"subtrackUrls Experiment={accUrl} Files={DOWNLOAD.format(db=db, track=track)}", "defaultSortField Cell_line", "maxCheckboxes 50", "noInherit on", "visibility hide", f"priority {priority}", ]) def strandOverlay(db, composite, exp, dataType, subDir, fileTag, shortLabel, longLabel, childLabel, priority): """One cell line's two strands as an overlay, plus strand drawn up and minus strand drawn down. The ProCapNet minus-strand files hold positive values and are flipped with negateValues; the ENCODE minus-strand signal is already negative and is not flipped. fileTag is the part of the file name that says what the file holds, so a bigWig downloaded on its own still names its source: the model for the predictions, the ENCODE experiment accession for the measurements. The parent line says "on", which makes these default-on and is what checks the signal data type in the faceted UI. The contribution scores stay off, so opening the composite gives signal rather than every data type at once.""" gbdb = GBDB.format(db=db, track=composite) + subDir negate = ["negateValues on"] if dataType == "pred" else [] out = stanza(8, [ f"track {composite}_{exp.cell}_{dataType}", f"parent {composite} on", "container multiWig", "aggregate solidOverlay", "showSubtrackColorOnUi on", "type bigWig", "autoScale on", "alwaysZero on", "maxHeightPixels 100:40:8", f"color {exp.color}", f"shortLabel {shortLabel}", f"longLabel {longLabel}", "onlyVisibility full", f"priority {priority}", ]) out += stanza(12, [ f"track {composite}_{exp.cell}_{dataType}_pos", f"parent {composite}_{exp.cell}_{dataType}", "type bigWig", f"bigDataUrl {gbdb}/{exp.cell}.{fileTag}.pos.bw", f"color {exp.color}", f"shortLabel {shortLabel} +", f"longLabel {childLabel}, plus strand", ]) out += stanza(12, [ f"track {composite}_{exp.cell}_{dataType}_neg", f"parent {composite}_{exp.cell}_{dataType}", "type bigWig", f"bigDataUrl {gbdb}/{exp.cell}.{fileTag}.neg.bw", f"color {exp.color}", f"altColor {exp.color}", ] + negate + [ f"shortLabel {shortLabel} -", f"longLabel {childLabel}, minus strand", ]) return out def contribStanza(db, exp, priority): gbdb = GBDB.format(db=db, track="proCapNet") return stanza(8, [ f"track proCapNet_{exp.cell}_contrib", "parent proCapNet off", "type bigWig", f"bigDataUrl {gbdb}/contrib/{exp.cell}.proCapNet-contrib.bw", "logo on", "autoScale on", "alwaysZero on", "maxHeightPixels 100:40:8", f"color {exp.color}", f"shortLabel {exp.cell} Contrib", f"longLabel {exp.cell} ProCapNet sequence-contribution scores", "onlyVisibility full", f"priority {priority}", ]) def proCapNetComposite(db, experiments): dataTypes = 'pred|"Predicted PRO-cap"' if db == "hg38": dataTypes += ' contrib|"Sequence contribution scores"' out = compositeStanza(db, "proCapNet", "ProCapNet", - "ProCapNet predicted PRO-cap and sequence-contribution scores", + "ProCapNet predicted PRO-cap", dataTypes, "https://www.encodeproject.org/annotations/$$/", 2) for priority, exp in enumerate(experiments, 1): out += strandOverlay(db, "proCapNet", exp, "pred", "/pred", "proCapNet", f"{exp.cell} Pred", f"{exp.cell} ProCapNet predicted PRO-cap, plus strand up and minus strand down", f"{exp.cell} ProCapNet predicted PRO-cap", priority) if db == "hg38": for priority, exp in enumerate(experiments, 11): out += contribStanza(db, exp, priority) return out def encode4ProCapComposite(db, experiments): out = compositeStanza(db, "encode4ProCap", "PRO-cap", "PRO-cap nascent RNA transcription start sites from ENCODE 4", 'procap|"PRO-cap"', "https://www.encodeproject.org/experiments/$$/", 1) for priority, exp in enumerate(experiments, 1): out += strandOverlay(db, "encode4ProCap", exp, "procap", "", exp.procapAcc, f"{exp.cell} PRO-cap", f"{exp.cell} PRO-cap transcription start sites, plus strand up and minus strand down", f"{exp.cell} PRO-cap transcription start sites", priority) return out def proCapNetTrackDb(opts, args): experiments = list(TsvReader(args.experimentsTsv)) fileOps.ensureFileDir(args.outRa) with fileOps.AtomicFileOpen(args.outRa) as fh: print(f"# Generated by hg/makeDb/outside/proCapNet/proCapNetTrackDb for {args.db}.", file=fh) print("# Do not edit by hand, edit the script and regenerate.\n", file=fh) fh.write(superStanza()) if args.db == "hg38": fh.write(encode4ProCapComposite(args.db, experiments)) fh.write(proCapNetComposite(args.db, experiments)) writeMetadata(f"{args.gbdbDir}/proCapNet/metadata.tsv", experiments, "modelAcc", args.db, "proCapNet") writeColors(f"{args.gbdbDir}/proCapNet/colors.json") if args.db == "hg38": writeMetadata(f"{args.gbdbDir}/encode4ProCap/metadata.tsv", experiments, "procapAcc", args.db, "encode4ProCap") writeColors(f"{args.gbdbDir}/encode4ProCap/colors.json") def main(): opts, args = parseArgs() with cli.ErrorHandler(noStackExcepts=(OSError, cli.PycbioException)): proCapNetTrackDb(opts, args) main()