PALM
Predict sequence aggregation profiles.
Results
Loading starting example…
Method & evidence
Apply the five-fold PALM ensemble to protein sequences and retain sequence scores, residue profiles, uncertainty, and candidate regions.
Scope & limitations
- The residue threshold is a screening choice rather than a calibrated cross-dataset decision boundary.
Interpreting results
- Scores are model predictions, not experimental aggregation rates.
- Residue scores are relative contributions inferred from sequence-level training.
- Compare candidates within a consistent model variant and threshold, then confirm experimentally.
References
Input requirements & outputs
Input checks
- Canonical 20-amino-acid alphabet
- At most 32 rows and 4,096 residues per job
- At most 1,022 residues per sequence
Expected results
- Sequence and residue profiles — Five-fold sequence scores, mean residue scores, uncertainty, and candidate regions.
- Immutable native evidence — Exact staged-asset, upstream, request, and result identity.
