94ca641d99a7745058c8df8a3c38309898cb5973 jnavarr5 Fri Sep 25 13:49:11 2026 -0700 Updating the script that generates the ra file for EpiCentral so it uses semicolons instead removing the commas outright. refs #38112 diff --git src/hg/makeDb/scripts/episignatures/epigenCentralToBed.py src/hg/makeDb/scripts/episignatures/epigenCentralToBed.py index 955d57e7052..7a9f45be260 100755 --- src/hg/makeDb/scripts/episignatures/epigenCentralToBed.py +++ src/hg/makeDb/scripts/episignatures/epigenCentralToBed.py @@ -1,262 +1,263 @@ #!/usr/bin/env python3 """Convert the EpigenCentral track hub bigBed into the bed we load as a native track. Reads the 16-column bed that bigBedToBed writes for https://github.com/ccmbioinfo/EpigenCentral-UCSC-Genome-Browser/blob/main/episignatures.bb and writes a 17-column bed, sorted, for bedToBigBed with epigenCentral.as. Four things change: - the OMIM column holds a full https://omim.org/entry/NNNNNN URL. Only the number is kept, so trackDb can make the link with "urls displayOmim=". - the hub's last-but-one column is a pre-rendered mouse-over, "NSD1|Loss|-0.346". It is replaced by just the direction, and trackDb builds the mouse-over from the fields. The direction is taken from the comparison table rather than by splitting the string, so it cannot drift from the table the user is shown. - rows of the comparison table are de-duplicated. 137 probes in the hub repeat a signature two or three times with identical values, while the sigCount and signatureList columns already count it once. - a displayDisorder column is added, read out of the comparison table, so the mouse-over can name the disorder and not only the gene. The comparison table is the ";" rows / "|" cells encoding that detailsDynamicTable expands inside a char[4096] in hgc.c, so the widest row is reported and checked. --raOut and --htmlOut write the filterValues.signatureList line for episignatures.ra and the "Included episignatures" table for epigenCentral.html straight from the data, so neither can drift away from the bigBed the way a hand-edited list would. """ import argparse import os import sys from collections import OrderedDict, Counter # The upstream bed columns, 0-based. OMIM_COL = 11 COUNT_COL = 12 LIST_COL = 13 MOUSEOVER_COL = 14 TABLE_COL = 15 N_COLS = 16 OMIM_PREFIX = "https://omim.org/entry/" # printEmbeddedTable() in hg/hgc/hgc.c expands the encoded table in a fixed buffer. HGC_TABLE_LIMIT = 4096 # Cells of a comparison-table row. GENE, DISORDER, OMIM, DIRECTION, DELTA, PVAL, CORRECTION = range(7) def parseTable(encoded, probe): """Split the ';'/'|' encoded table into a header and a list of rows.""" parts = encoded.split(";") header = parts[0].split("|") rows = [] for part in parts[1:]: cells = part.split("|") if len(cells) != len(header): raise ValueError("%s: comparison row %r has %d cells, header has %d" % (probe, part, len(cells), len(header))) rows.append(cells) return header, rows def dedupeRows(rows): """Drop repeated rows, keeping the first occurrence and the original order.""" seen = OrderedDict() for cells in rows: seen.setdefault("|".join(cells), cells) return list(seen.values()) def readRefs(fname): """signature -> (pmid, doi) from the checked-in reference table.""" refs = {} with open(fname) as fh: for line in fh: if line.startswith("#") or not line.strip(): continue cells = line.rstrip("\n").split("\t") cells += [""] * (3 - len(cells)) refs[cells[0]] = (cells[1].strip(), cells[2].strip()) return refs def loadAuthorYear(): """id (PMID or DOI) -> "Lastname Year", from the checked-in lookup shared with makeHtmlTables.py.""" fname = os.path.join(os.path.dirname(os.path.abspath(__file__)), "pubAuthorYear.tsv") authorYear = {} with open(fname) as fh: for line in fh: if line.startswith("#") or not line.strip(): continue pmid, ay = line.rstrip("\n").split("\t") authorYear[pmid] = ay return authorYear def writeRa(fh, signatures): """The filterValues/filterType lines for the signature menu. The menu label is "Disorder (SIGNATURE)". A comma in a filterValues entry is the entry separator and, with a *List* filterType, cannot be escaped at all, so the - commas inside a few disorder names are dropped here rather than in the bigBed. + commas inside a few disorder names become semicolons here rather than in the + bigBed, as the methaDory menu does. """ entries = [] for sig in sorted(signatures, key=str.lower): - disorder = signatures[sig]["disorder"].replace(",", "") + disorder = signatures[sig]["disorder"].replace(",", ";") entries.append("%s|%s (%s)" % (sig, disorder, sig)) fh.write(" filterValues.signatureList %s\n" % ",".join(entries)) fh.write(" filterType.signatureList multipleListAnd\n") def writeHtml(fh, signatures, refs): """The "Included episignatures" table of the description page.""" authorYear = loadAuthorYear() fh.write('\n') fh.write("" "\n") for sig in sorted(signatures, key=str.lower): rec = signatures[sig] pmid, doi = refs.get(sig, ("", "")) id_ = pmid or doi if not id_: raise ValueError("no reference for signature %s in the reference table" % sig) if id_ not in authorYear: raise ValueError("no author/year in pubAuthorYear.tsv for id %s" % id_) if pmid: ref = ('%s' % (pmid, authorYear[id_])) else: ref = ('%s' % (doi, authorYear[id_])) fh.write("" '' "\n" % (sig, rec["disorder"], rec["omim"], rec["omim"], rec["probes"], ref)) fh.write("
EpisignatureDisorderOMIMCpG probesReference
%s%s%s%d%s
\n") def main(): parser = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter) parser.add_argument("--raOut", help="write the signature filter menu for episignatures.ra") parser.add_argument("--htmlOut", help="write the episignature table for epigenCentral.html") parser.add_argument("--refs", help="signature to PMID/DOI table, needed for --htmlOut") args = parser.parse_args() dropped = 0 # duplicate comparison rows removed dedupedProbes = 0 # probes that had at least one naDeltas = 0 # rows the hub has no delta-beta for widest = 0 widestProbe = None out = [] signatures = {} # signature -> disorder, omim, probe count directions = Counter() for lineNo, line in enumerate(sys.stdin, start=1): row = line.rstrip("\n").split("\t") if len(row) != N_COLS: raise ValueError("line %d: expected %d columns, found %d" % (lineNo, N_COLS, len(row))) probe = row[3] header, rows = parseTable(row[TABLE_COL], probe) nRows = len(rows) rows = dedupeRows(rows) if len(rows) != nRows: dropped += nRows - len(rows) dedupedProbes += 1 # The columns the hub derives from the table have to agree with it. listed = sorted(row[LIST_COL].split(",")) inTable = sorted(cells[GENE] for cells in rows) if listed != inTable: raise ValueError("%s: signatureList %s does not match the comparison " "table %s" % (probe, listed, inTable)) if int(row[COUNT_COL]) != len(rows): raise ValueError("%s: sigCount is %s but the comparison table has %d rows" % (probe, row[COUNT_COL], len(rows))) naDeltas += sum(1 for cells in rows if cells[DELTA] == "NA") for cells in rows: rec = signatures.setdefault(cells[GENE], {"disorder": cells[DISORDER], "omim": cells[OMIM], "probes": 0}) if rec["disorder"] != cells[DISORDER] or rec["omim"] != cells[OMIM]: raise ValueError("%s: signature %s is %r/%s here but %r/%s elsewhere" % (probe, cells[GENE], cells[DISORDER], cells[OMIM], rec["disorder"], rec["omim"])) rec["probes"] += 1 # The strongest signature, as the hub picked it, gives the color, the # direction and the disorder shown on the mouse-over. display = [cells for cells in rows if cells[GENE] == row[10]] if not display: raise ValueError("%s: displayed signature %r is not in the comparison table" % (probe, row[10])) display = display[0] if display[DELTA] == "NA": raise ValueError("%s: displayed signature %r has no delta-beta" % (probe, row[10])) if abs(abs(float(display[DELTA])) - float(row[9])) > 1e-9: raise ValueError("%s: maxAbsDelta %s is not |%s|" % (probe, row[9], display[DELTA])) omim = row[OMIM_COL] if omim.startswith(OMIM_PREFIX): omim = omim[len(OMIM_PREFIX):].strip("/") if omim != display[OMIM]: raise ValueError("%s: OMIM column %r disagrees with the comparison table %r" % (probe, omim, display[OMIM])) encoded = ";".join(["|".join(header)] + ["|".join(cells) for cells in rows]) if len(encoded) > widest: widest, widestProbe = len(encoded), probe out.append(row[0:10] + [ row[10], # displaySignature display[DISORDER], # displayDisorder omim, # displayOmim display[DIRECTION], # direction row[COUNT_COL], # sigCount row[LIST_COL], # signatureList encoded, # comparisonTable ]) directions[display[DIRECTION]] += 1 out.sort(key=lambda r: (r[0], int(r[1]))) for row in out: sys.stdout.write("\t".join(row) + "\n") if args.raOut: with open(args.raOut, "w") as fh: writeRa(fh, signatures) if args.htmlOut: if not args.refs: raise SystemExit("--htmlOut needs --refs") with open(args.htmlOut, "w") as fh: writeHtml(fh, signatures, readRefs(args.refs)) sys.stderr.write("features written: %d\n" % len(out)) sys.stderr.write("episignatures: %d\n" % len(signatures)) sys.stderr.write("displayed direction: %s\n" % ", ".join("%s %d" % kv for kv in sorted(directions.items()))) sys.stderr.write("duplicate table rows dropped: %d, on %d probes\n" % (dropped, dedupedProbes)) sys.stderr.write("table rows with no delta-beta: %d\n" % naDeltas) sys.stderr.write("widest comparison table: %d bytes (%s), hgc limit %d\n" % (widest, widestProbe, HGC_TABLE_LIMIT)) if widest >= HGC_TABLE_LIMIT: raise SystemExit("ERROR: a comparison table is too wide for hgc.c, it would " "errAbort on the details page. Switch the field to the " "_json encoding, see methaDoryToBed.py.") if __name__ == "__main__": main()