0f951bfc96241a7362d1512ceecf59a02682a839
markd
  Thu Sep 24 20:40:52 2026 -0700
Description page fixes from a QA pre-pass on the TSS tracks. refs #35528

Encode the non-ASCII character in the Avsec reference on proCapNet.html.
getTrackReferences emits raw UTF-8, which the browser does not transcode.

Use $db rather than a hardcoded hg38 in the proCapNet download-server link and
the bigWigToBedGraph example. One page serves both assemblies, so an hs1 reader
was being pointed at hg38 files.

Add a Source subsection to proCapNet.html linking the makedoc, the build scripts,
the trackDb file and the upstream kundajelab/ProCapNet repository. These links
went missing when the Processing at UCSC section was dropped; the prose stays
dropped.

Drop the cross-links between the two track pages. Track names are prefixed
hub_<id>_ on hs1, which is a curated hub, and the id is machine-specific, so a
bare-name link cannot work there. Each page now states which assemblies the data
is available on instead, naming both GRCh38/hg38 and T2T-CHM13/hs1.

Bold the two UI control names on the proCapNet display conventions section.

diff --git src/hg/makeDb/trackDb/human/transcriptionStart.html src/hg/makeDb/trackDb/human/transcriptionStart.html
index 5506f6625a7..88bf531564e 100644
--- src/hg/makeDb/trackDb/human/transcriptionStart.html
+++ src/hg/makeDb/trackDb/human/transcriptionStart.html
@@ -22,44 +22,45 @@
 <p>
 No single assay settles where a TSS is. Run-on methods catch the nascent RNA at
 the moment it is made, cap-based methods read the protected 5' end of a finished
 transcript, long reads follow a transcript from one end to the other, and
 sequence models predict initiation without an experiment at all. Each sees a
 different slice of the same event, and they disagree in informative ways. This
 collection gathers those lines of evidence in one place so they can be compared
 at a locus.
 </p>
 
 <p>
 Tracks currently in the collection:
 </p>
 
 <ul>
-<li><a href="hgTrackUi?g=encode4ProCap">PRO-cap</a>: initiation events measured
-directly by PRO-cap in six cell lines by ENCODE 4.</li>
-<li><a href="hgTrackUi?g=proCapNet">ProCapNet</a>: genome-wide, base-resolution
+<li>PRO-cap: initiation events measured
+directly by PRO-cap in six cell lines by ENCODE 4. Available on GRCh38/hg38
+only.</li>
+<li>ProCapNet: genome-wide, base-resolution
 predictions of initiation from DNA sequence alone, from six models each trained
 on one of those cell lines, together with per-base scores showing which bases
-each model used.</li>
+each model used. The predictions are available on GRCh38/hg38 and
+T2T-CHM13/hs1; the per-base scores are available on GRCh38/hg38 only.</li>
 </ul>
 
 <p>
 Each track has its own description page covering what it measures or predicts,
-how it was made, and how to download it. On T2T-CHM13 only the ProCapNet
-predictions are available, since the PRO-cap experiments and the contribution
-scores were produced against GRCh38. More TSS evidence tracks will be added here
-over time.
+how it was made, and how to download it. The PRO-cap experiments and the
+contribution scores were produced against GRCh38, which is why neither is on
+T2T-CHM13. More TSS evidence tracks will be added to this collection over time.
 </p>
 
 <h2>Display Conventions and Configuration</h2>
 
 <p>
 Tracks are grouped by the evidence they carry, with measurements before
 predictions. Signal tracks that distinguish the two strands draw the plus strand
 upward and the minus strand downward, so a divergent promoter reads as a pair of
 peaks straddling the element.
 </p>
 
 <h2>Data Access</h2>
 
 <p>
 Each track listed above has its own description page with details on methods,