KGIPA

KGIPA's online predictor

Enter the query peptide and protein sequences in FASTA format:

Select model: KGIPA KGIPA-Fast

Perform SHAP explainability analysis (this will increase computation time; select only if needed)

Email (optional, you will receive a notification when the analysis is finished):

Example Reset

Note:
1️⃣ KGIPA — the full version proposed in our study. It integrates multiple features including sequence encoding, physicochemical properties, secondary structure, intrinsic disorder tendency, evolutionary information, pretrained embeddings, and trRosetta-predicted 3D structural features. This version achieves high prediction accuracy but requires longer computation time (≈10 min).
2️⃣ KGIPA-Fast — a compact variant of KGIPA that removes the 3D structural features derived from trRosetta and the secondary structural features obtained from SCRATCH, sacrificing a small amount of accuracy while greatly improving computational efficiency (≈2 min).
3️⃣ SHAP analysis — providing peptide- and protein-level feature importance along with complete feature and contribution details. Executing SHAP analysis typically increases runtime by about 20 additional minutes (depending on GPU load).
4️⃣ For high-throughput analyses, we recommend downloading and running the original implementation from our GitHub repository ( https://github.com/ShutaoChen97/KGIPA ) locally for optimal efficiency.