8296c9b7908b7fd73c70b6c7af21c337c55558f4 lrnassar Mon Sep 21 04:39:42 2026 -0700 Wording and formatting fixes to the mouseStrainsCactus description page and the AVI news entry, per CR feedback. refs #38351 mouseStrainsCactus.html: use the standard "Display Conventions and Configuration" heading, spell misassemblies without the hyphen, hyphenate large-scale, add the missing commas before "and" in two sentences, finish the Data Access sentence about the API track name, and put the references in alphabetical order by first author. newsarch.html: correct the track group name to "Phenotypes, Variants, and Literature", hyphenate genome-wide, and add the missing "on" in the figure caption. diff --git src/hg/htdocs/goldenPath/newsarch.html src/hg/htdocs/goldenPath/newsarch.html index 8fe053fbf81..7c0a1336af5 100644 --- src/hg/htdocs/goldenPath/newsarch.html +++ src/hg/htdocs/goldenPath/newsarch.html @@ -72,44 +72,44 @@ (AVI) score from Google DeepMind's AlphaGenome Atlas, on the human GRCh38/hg38 assembly. The score combines AlphaGenome's predictions of how a variant affects gene regulation, covering expression, splicing, chromatin accessibility and transcription factor binding across hundreds of cell types, with AlphaMissense predictions for protein-altering changes. The result is a single number that ranks how damaging a substitution is likely to be. Unlike most prediction scores, it covers non-coding variants as well as coding ones, across the whole genome.

AlphaGenome AVI scores across the TERT locus, showing four per-allele subtracks with
 elevated scores over the gene's 5' end and its highlighted promoter

AVI scores at the TERT locus on hg38, one subtrack per alternate allele. Scores stay high across the highlighted promoter, shown in red as EH38E3622530, and fall -to background either side. The two promoter mutations most often seen in cancer sit inside it, +to background on either side. The two promoter mutations most often seen in cancer sit inside it, beyond the end of every TERT transcript.

The track shows the precomputed score for every possible single-base substitution, about 8.8 billion in all, with one subtrack per alternate allele. Scores are PHRED-scaled, so 10 marks the -top 10 percent of substitutions genome wide, 20 the top 1 percent and 30 the top 0.1 percent. +top 10 percent of substitutions genome-wide, 20 the top 1 percent and 30 the top 0.1 percent. Hover the mouse over any position to see the score, and zoom in to see individual bases.

-The AlphaGenome Variant Impact (AVI) score track can be found in the "Phenotype and Literature" -group, inside the +The AlphaGenome Variant Impact (AVI) score track can be found in the +"Phenotypes, Variants, and Literature" group, inside the Deleteriousness Predictions track collection, so it can be compared with other prediction scores such as CADD, REVEL, EVE, PrimateAI and PromoterAI. AlphaGenome has not been validated for, and is not approved for, any clinical use.

We thank Google DeepMind for making the data available. See the track description page for details on the data, methods and licensing. The AVI data cannot be redistributed by UCSC, so it