Each item carries the assay score, the functional class that score falls into, and, where the
score set has been calibrated against variants of known clinical significance, the ACMG/AMP
functional evidence code that follows: PS3 for evidence of a damaging effect, BS3 for evidence
of a normal one, at a strength set by the calibration's odds of pathogenicity following the
-ClinGen recommendations for the PS3/BS3 criterion (Brnich et al.). ClinVar
+ClinGen recommendations for the PS3/BS3 criterion (Brnich et al., 2019). ClinVar
significance, gnomAD allele frequency and the ClinGen allele ID are attached where they exist.
Each item also carries the assay metadata: what the assay measured, in what model system, the
molecular mechanism assessed, and how the variant library was built. That last field matters
clinically. An assay built on an in vitro construct library introduces a synthetic copy of the
target sequence, so it cannot detect an effect on splicing or on nonsense-mediated decay and
can read falsely normal for such a variant. An assay that edits the endogenous locus can.
Two cautions when comparing variants. Many of the evidence codes come from calibrations MaveDB
marks research use only; each item names its calibration and flags this. And the functional
score has no common scale or direction between score sets, so a higher score does not always
@@ -35,32 +35,32 @@
Items are colored by the ACMG functional evidence code from the calibration named in the item
details. Red shades carry evidence toward pathogenic, blue toward benign, darkening with the
strength of the evidence.
Score sets that classify their variants without assigning an ACMG code are a measurement rather
than a clinical claim, and use a separate palette:
| |
Abnormal function measured, no ACMG evidence code assigned |
|---|
@@ -73,31 +73,31 @@
Filters are available on the ACMG evidence code, the functional class, the gene, the ClinVar
significance and the assay metadata, and none are applied by default. Because every
substitution at a codon is a separate item, a well studied gene stacks deeply and the track
falls back to a density graph in all but narrow windows; filtering to one gene or one evidence
code makes individual variants readable. Item names, ClinGen allele IDs and MaveDB variant
identifiers are searchable from the position box.
Methods
MaveMD curates MaveDB score sets for clinical relevance, integrates them with ClinVar and the
ClinGen Allele Registry, and exports evidence structured for ACMG/AMP variant classification
-(McEwen et al.). Candidates were drawn from MaveDB, from publications cited in ClinVar
+(McEwen et al., 2025). Candidates were drawn from MaveDB, from publications cited in ClinVar
for functional evidence, and from community recommendation, then restricted to genes with a
moderate or stronger gene-disease association in ClinGen or GenCC.
Data are downloaded from the MaveDB API,
one request per score set. MaveDB resolves some score sets to the genome and others only to a
protein sequence, so a variant is placed from its genomic term where one exists, otherwise by
projecting its protein term onto the corresponding codon through RefSeq or GENCODE, otherwise
by converting the submitter's transcript term. Haplotypes, several substitutions measured as a
single unit, have no one position and are excluded.
The build scripts are in the
@@ -131,57 +131,56 @@
-start=7668402 -end=7687550 stdout
The original annotation source data can be downloaded from the MaveDB API at
https://api.mavedb.org.
Credits
Thanks to Alan Rubin, Benjamin Capodanno and Jeremy Stone of the MaveDB team for curating the
MaveMD collection and for advice on retrieving it, and to the many groups whose experiments
make up the underlying score sets.
References
-
-McEwen AE, Stone J, Tejura M, Gupta P, Capodanno BJ, Da EY, Grindstaff SB, Moore N, Reinhart D,
-Snyder AE et al.
-
-MaveMD: A functional data resource for genomic medicine.
-medRxiv. 2025 Nov 19;.
-PMID: 41332838; PMC: PMC12668102
-
-
-
-Rubin AF, Stone J, Bianchi AH, Capodanno BJ, Da EY, Dias M, Esposito D, Frazer J, Fu Y, Grindstaff
-SB et al.
-
-MaveDB 2024: a curated community database with over seven million variant effects from multiplexed
-functional assays.
-Genome Biol. 2025 Jan 21;26(1):13.
-PMID: 39838450; PMC: PMC11753097
-
-
Arbesfeld JA, Da EY, Stevenson JS, Kuzma K, Paul A, Farris T, Capodanno BJ, Grindstaff SB, Riehle K,
Saraiva-Agostinho N et al.
Mapping MAVE data for use in human genomics applications.
Genome Biol. 2025 Jun 25;26(1):179.
PMID: 40563119; PMC: PMC12188674
Brnich SE, Abou Tayoun AN, Couch FJ, Cutting GR, Greenblatt MS, Heinen CD, Kanavy DM, Luo X, McNulty
SM, Starita LM et al.
Recommendations for application of the functional evidence PS3/BS3 criterion using the ACMG/AMP
sequence variant interpretation framework.
Genome Med. 2019 Dec 31;12(1):3.
PMID: 31892348; PMC: PMC6938631
+
+McEwen AE, Stone J, Tejura M, Gupta P, Capodanno BJ, Da EY, Grindstaff SB, Moore N, Reinhart D,
+Snyder AE et al.
+
+MaveMD: A functional data resource for genomic medicine.
+medRxiv. 2025 Nov 19;.
+PMID: 41332838; PMC: PMC12668102
+
+
+
+Rubin AF, Stone J, Bianchi AH, Capodanno BJ, Da EY, Dias M, Esposito D, Frazer J, Fu Y, Grindstaff
+SB et al.
+
+MaveDB 2024: a curated community database with over seven million variant effects from multiplexed
+functional assays.
+Genome Biol. 2025 Jan 21;26(1):13.
+PMID: 39838450; PMC: PMC11753097
+