AiPP: Artificial Intelligence Protein Profiling

Webmaster: G.W. Dayhoff II

About AiPP

AiPP illuminates reversible and covalently ligandable sites across the human proteome, including drug targets undetectable by chemical proteomic approaches.

Of the 19,486 unique proteins in the human proteome (UniProt database https://www.uniprot.org/), 9,580 have not been detected and 3,362 have not been liganded by ABPP (as in the LigCysABPP database). A total of 1,860 proteins (1,472 ABPP liganded and 388 ABPP unliganded) are in the model training set. Drug targets within ABPP undetected or unliganded proteins are stratified by functional family classifications.

How to use

  • Explore — search the AiPP Atlas by UniProt ID or gene symbol.
  • Inspect — view curated experimental evidence from LigCysABPP.
  • Predict — score any sequence not in the Atlas (ESM Forge token required).
Figure panels showing human proteome coverage and Top-1 score distributions for drug targets unseen by ABPP

🗺️ Explore

Explore the AiPP Atlas: a comprehensive resource of precomputed site-specific predictions for human proteins.

(note: proteins comprising 2046 or fewer residues)

🔬 Inspect

Inspect the LigCysABPP database: a curated repository of experimental evidence for human cysteines.

(note: search by UniProt ID or Gene Symbol)

🤖 Predict

Predict reversible and covalently ligandable sites in any sequence using AiPP: non-human, custom constructs, or variants.

(note: an ESM Forge Token is required)