Sequence variation, splicing and gene regulation. Explore models we could test and the test families that apply to them.
These are candidates, not a ranked list. An applicable test may need a trained head, a scoring adapter, a specialist comparison, or access to a hosted service. No result is implied by appearing here.
Public checkpoint; remote model code needs review before local use.
Official source ↗Paper-reported results →Reference/mutant embeddings need a head fitted on the MFASS training split.
Sequence representation needs assay-specific calibration.
Public checkpoint.
Official source ↗Paper-reported results →Paired variant representation needs a train-only head.
Use a head fitted without held-out assay leakage.
Public checkpoints; official local inference needs CUDA hardware and substantial memory.
Official source ↗Paper-reported results →Likelihood or embeddings need a predeclared variant-effect adapter.
Long context is available, but output needs a task-specific mapping.
Rate-limited, non-commercial API requires a key. Downloadable weights require accepting non-commercial model terms; local inference recommends an H100 GPU.
Official source ↗Paper-reported results →Regulatory outputs require a declared MPRA scoring adapter; access is through the API or downloadable weights under model terms.
Native long-context regulatory outputs via the API or downloadable weights; local inference needs substantial GPU resources.
Public archived code under PolyForm Strict; model weights are CC BY-NC 4.0 for non-commercial use.
Official source ↗Paper-reported results →Native splice-effect score is an existing specialist comparator.
Public specialist code and models under GPL-3.0.
Official source ↗Paper-reported results →Native splice-effect score is an existing specialist comparator.
Each candidate test needs a defined dataset, split, metric and run protocol before it can become a benchmark.
Functional exon-recognition assay; mfass-v2 reports a corrected baseline and one complete local DNABERT-2 protocol.
Read the corrected MFASS v2 result →Predict measured activity changes from regulatory sequence variants.
Predict gene-expression or chromatin effects from long-context DNA sequence.
Model sources and access descriptions checked 2026-09-16. Confirm licenses, terms and checkpoint revisions before any run.