The Kinase Atlas: KinoPlex Maps Phosphorylation Across the Human Proteome
KinoPlex maps ~250,000 high-confidence phosphorylation candidates across the human proteome using AlphaFold structures and kinase motifs — published in Nature Biotechnology July 29.
Of the 1.8 million serine, threonine, and tyrosine residues in the human proteome, only about 6 percent bear experimental validation of phosphorylation. A paper published July 29, 2026, in Nature Biotechnology introduces KinoPlex, integrating predicted protein structures and kinase recognition motifs to assign phosphorylation potential across the proteome.
Researchers used roughly 20,000 AlphaFold models and positive-unlabeled transfer learning to identify about 567,000 structurally phospho-competent residues, yielding roughly 250,000 high-confidence candidates with both sequence recognition potential and favorable structural presentation.
Sequence–structure selective coupling
The atlas uncovered organizing principles the authors call sequence–structure selective coupling: kinases achieve specificity through structural scarcity or accessibility of preferred motifs — not motif discrimination alone. Deep phosphoproteomics in K562 cells validated predictions. Data is available at kinoplex.phosphosite.org.
Why it matters
Incomplete substrate maps slow kinase drug programs. KinoPlex turns AlphaFold structures into a searchable atlas of hypotheses — a pattern biotech teams increasingly expect from AI-native discovery stacks.
Predictions remain computational until wet-lab confirmation catches up, but the release is among the largest open resources tying structural biochemistry to proteome-wide kinase search.
Sources
- Nature Biotechnology — An AI-enabled structural atlas decodes kinase specificity across the human proteome (July 29, 2026)
- KinoPlex — Structural Atlas of Human Phosphorylation (2026)
- Zenodo — KinoPlex structural atlas datasets (2026)