CodeXome has played an important part in the examination of variants of uncertain significance. I successfully identified variants shared with other primates, which enabled their instant classification as benign.
Every other tool predicts. We show you what selection observed.
CodeXome is the only variant evidence layer built on direct observation, not statistical inference. For every coding residue, we report what 80 million years of primate selection has actually tolerated — or never permitted — across 55 genera.
After standard filtering, hundreds of plausible candidates remain. Most are biologically tolerated, but you still have to look at them.
Standard pipeline output
~300
candidate missense variants per exome
After gnomAD frequency filtering and ML predictor scoring, hundreds of plausible candidates remain — 75–150 hours of manual review per case before functional work can begin.
With CodeXome annotation
40–75
variants worth your follow-up time
50–75% reduction by removing variants that recur across primate evolution. Selection has already tested them, and they passed. Recall on ClinGen-classified pathogenic variants: preserved.
Across 52 ClinGen expert-curated disease genes, no expert-classified pathogenic variant co-occurred with primates in the database.
Where the evidence comes from
An evolutionary window between population frequency and deep conservation.
Population frequency tools see the last few thousand years. Deep vertebrate conservation sees hundreds of millions. Primate constraint occupies the intermediate window — close enough to humans for biochemical relevance, broad enough to capture patterns that human cohorts alone cannot resolve.
Last 10,000 years
Human population
Variation observed in modern human cohorts. Captures recent allele frequencies but a narrow time window.
ExamplesgnomAD · LOEUF · ClinVar
Where CodeXome lives ~80 million years
Primate evolutionary constraint
Empirical recurrence across 55 primate genera, mapped to human GRCh38 coordinates at residue-level resolution.
Coverage19,244 genes
100 million – 1 billion years
Deep vertebrate conservation
Conservation across hundreds of millions of years. Sensitive at deeply conserved sites; multiple hits per site at long timescales reduce resolution.
ExamplesphyloP · phastCons · GERP++
Inside the primate window, gene change tracks well-resolved speciation — the linear region of the evolutionary clock, where each substitution carries phylogenetic signal. Move further out and multiple hits per site begin to erase that signal. Move closer and the sampling collapses. The intermediate window is where empirical recurrence remains both biologically interpretable and statistically informative.
codexome gene previewer
Try it on a gene.
Enter a gene symbol see a glimpse of the alignment across 55 primate genera, mapped to human GRCh38 coordinates.
Run a Gene✓
Cross-primate alignment
55 genera, residue-level resolution
Free
Unlocked with platform access
🔒 Residue-level constraint
🔒 DTAS scoring
🔒 VCF annotation
🔒 TSV export
🔒 Cohort-scale processing
🔒 Gene Profile Module
codexome studies
All studies →Benchmarked across rare disease, hereditary cancer, and ClinGen-curated panels.
BRCA Exchange benchmark
100%
Concordance with expert panels
Pathogenic variants in BRCA1/BRCA2 expert-classified by the BRCA Exchange were unique to humans in the primate database. Benign variants were broadly shared.
ClinGen 52-gene concordance
52/52
Genes with full concordance
Across 52 ClinGen expert-curated disease genes, every pathogenic variant was human-specific in the primate database — no expert-classified pathogenic variant co-occurred with primates.
CFTR therapeutic target benchmark
99.3%
Accuracy on 260 drug-target mutations
CodeXome scoring matched 99.3% of 260 mutations targeted by approved CFTR drug therapeutics — an external, biology-anchored benchmark of relevance to gene-targeted therapy programs.
14,000-gene VUS survey
~20%
Of VUS missense identified as likely benign
Across ~14,000 ClinVar genes, an average of 20% of VUS missense variants (range 10–40% per gene) were shared with other primates — identifiable as likely benign on evolutionary evidence alone.
How it fits your pipeline
Three steps, from VCF in to prioritized candidates out.
01 / Upload
VCF or gene list
GRCh38-aligned VCFs, gene symbols, Ensembl IDs, or cohort-scale uploads. No preprocessing required.
02 / Annotate
Primate recurrence + DTAS
Every variant annotated with cross-primate recurrence, residue-level constraint, and Deep Time Ancestry Score.
03 / Export
TSV, VCF, or Gene Profile
Export annotated variants as TSV or VCF, or pull full Gene Profile reports for structured review.
Why this exists
A primate biology resource, two decades in the making, applied to human disease.
CodeXome did not begin as a product. It began with a scientific observation: humans share approximately 20,000 protein-coding genes with their closest evolutionary relatives, and 80 million years of primate evolution contains a record of which changes nature has tolerated and which it has not.
The biological resource that makes this observation usable — one of the most comprehensive primate sample collections ever assembled by a public research institution — was preserved through the closure of its original NCI laboratory and relocated to a formal research facility under NSF Phase II funding. 55 genera, 19,244 genes. Assembled, curated, and released for researchers to use today.
.jpg)
Early access program
A small cohort of research labs, working with us first.
We're partnering with research groups working on rare disease, hereditary cancer, and gene-targeted therapeutics.
3 / 5slots remaining