Loading…
Whole-protein carbohydrate-binding classification from structure, for proteome-scale screening with graph convolutions over ESM2 embeddings.
No providers recorded yet. Browse all providers
How much of a proteome touches sugar? The question is hard to answer experimentally, because settling it would require testing every protein against a large glycan library. Current catalogs label only 1.5 to 5% of proteins as carbohydrate-binding, yet 50 to 70% of proteins are glycosylated — a mismatch suggesting that carbohydrate recognition is far more widespread than the annotated lectin repertoire implies.
PiCAP (Protein interaction of CArbohydrate Predictor), from the GrayLab at Johns Hopkins, attacks that gap computationally by making a whole-protein call: does this protein bind a carbohydrate non-covalently? It was released with CAPSIF2, which localizes the interaction to specific residues once a protein has been flagged; the two share a training corpus, an equivariant graph backbone, and a single inference entry point, and ship as one package with fixed weights applicable to any new structure.
Preprinted in March 2025 and published in PNAS, the work's headline result comes from running across three complete proteomes, where 35 to 40% of proteins are called carbohydrate-binding — an order of magnitude above prior estimates.
The network is built from equivariant graph convolutional layers operating on C-beta coordinates, with radial basis function edge features encoding inter-residue distance and orientation, and takes per-residue ESM2 embeddings as node inputs. Twelve residual layers at 128 embedding dimensions feed adaptive pooling to a fixed length of 150 and then a convolutional-plus-dense classification head, totaling 1,798,895 parameters.
The NoCAP dataset holds 30,849 unique structures: 9,608 with bound carbohydrates drawn from DIONYSUS, UniLectin, SAbDab, and the original CAPSIF set, against 21,242 curated non-binders. Sequences were clustered at 60% identity with MMseqs before an 80/5/15 split into 24,957 training, 1,479 validation, and 4,413 test structures. Held-out balanced accuracy is 89.6% — a 96.3% true positive rate against an 82.8% true negative rate — with perfect rejection of ribosomal proteins but weaker discrimination on designed proteins (60.8%) and antibodies (56.2%). Applied to whole proteomes, the model called 1,677 of 4,339 E. coli proteins (39%), 8,177 of 21,304 mouse proteins (38%), and 7,029 of 20,067 human proteins (35%), agreeing with 225 of 225 known mouse lectins and 229 of 230 human lectins from LectomeXplore.
Glycobiologists gain a way to nominate candidate carbohydrate-binding proteins before committing to glycan array or pull-down experiments. Because the pipeline consumes predicted structures, it scales to organisms with no experimental structural coverage, which is what makes a whole-proteome pass feasible at all. A positive call is the natural input to CAPSIF2, which turns the yes/no answer into a residue-level site map suitable as a docking restraint.
The substantive claim here is biological rather than architectural: if a third or more of a proteome engages carbohydrates, glycan recognition is a general property of protein surfaces rather than a specialty of a small lectin family. The authors are explicit that this first pass is noisy and requires experimental follow-up, and they identify concrete sources of error — the negative set is likely contaminated with undiscovered binders, crystallographic complexes over-represent high-affinity interactions relative to physiological avidity-driven binding, and performance degrades on antibodies and designed proteins whose hypervariable sequences are poorly represented in language model pretraining. Weights are distributed openly from the lab's server and the code is MIT-licensed, though the convenience Docker image bundles PyRosetta, which requires a commercial license for non-academic use.
Much of this page is generated or calculated automatically. Flag anything that looks off and we will re-run it.