Submit Sequences
Paste one sequence, multi-FASTA, or upload a CSV/Excel file for batch prediction.
Batch summary report
| # | Identifier | Length | P(LLPS) | Logit | Class |
|---|
Methods
A four-stage pipeline from raw sequence to calibrated LLPS probability.
The input sequence is validated, then embedded by the ESM2-650M transformer, producing a per-residue representation of dimension L × 1280. Masked mean and max pooling along the residue axis yields a fixed 2560-dim vector capturing both average and salient local features.
This vector passes through a small fully-connected classifier — 2560 → 128 → 1 with ReLU and dropout — trained to discriminate LLPS-associated nucleic-acid binding proteins from non-LLPS NABs. The output logit is converted to a probability via sigmoid, with a default threshold of τ = 0.50.
L ∈ [10, 2000], 20-AA.
shape L × 1280.
concatenated to ℝ²⁵⁶⁰.
threshold τ = 0.50.
API Reference
Single or batch JSON endpoint for scripted and programmatic use.
POST /predict accepts either { "sequence": "..." } for one sequence or { "sequences": [...] } for a batch (max 16 per request — chunk client-side for larger sets).
POST /predict{ "sequence": "MTEIT…YL" } → { classification, score, logit, threshold, length }{ "sequences": [{"id":"FUS","sequence":"…"}, …] } → { count, ok, positives, errors, results: [...] }GET /health → { status, hf_token_configured, max_batch }