Protein embeddings and deep learning predict binding residues for various ligand classes

Abstract

One important aspect of protein function is the binding of proteins to ligands, including small molecules, metal ions, and macromolecules such as DNA or RNA. Despite decades of experimental progress many binding sites remain obscure. Here, we proposed bindEmbed21, a method predicting whether a protein residue binds to metal ions, nucleic acids, or small molecules. The Artificial Intelligence (AI)-based method exclusively uses embeddings from the Transformer-based protein Language Model (pLM) ProtT5 as input. Using only single sequences without creating multiple sequence alignments (MSAs), bindEmbed21DL outperformed MSA-based predictions. Combination with homology-based inference increased performance to F1 = 48 ± 3% (95% CI) and MCC = 0.46 ± 0.04 when merging all three ligand classes into one. All results were confirmed by three independent data sets. Focusing on very reliably predicted residues could complement experimental evidence: For the 25% most strongly predicted binding residues, at least 73% were correctly predicted even when ignoring the problem of missing experimental annotations. The new method bindEmbed21 is fast, simple, and broadly applicable—neither using structure nor MSAs. Thereby, it found binding residues in over 42% of all human proteins not otherwise implied in binding and predicted about 6% of all residues as binding to metal ions, nucleic acids, or small molecules.

Contrastive learning on protein embeddings enlightens midnight zone

“…by using the output of the last hidden layers of the networks forming the pLMs, yields a representation of protein sequences referred to as embeddings (Figure 1 in ( 37 )). Embeddings have been used successfully as exclusive input to predicting secondary structure and subcellular localization at performance levels almost reaching ( 38–40 ) or even exceeding ( 37 , 46 , 47 ) the SOTA using evolutionary information from MSAs as input. Embeddings can even substitute sequence similarity for homology-based annotation transfer ( 48 , 49 ).…”

Structural Bioinformatics and Deep Learning of Metalloproteins: Recent Advances and Applications

“…The bindEmbed21 approach combines homology-based inference with ML to predict whether a protein residue binds to a metal ion, a nucleic acid, or a small molecule [ 95 ]. The ML component used protein embeddings as inputs to a two-layer convolutional neural network (CNN).…”

LambdaPP: Fast and accessible protein-specific phenotype predictions

“…predicting CATH (62) classes (25), LambdaPP will be updated to extend the breadth of pLM-based predictions offered. All feature prediction methods integrated into the LambdaPP webserver are currently based on ProtT5 embeddings (23) as methods trained on ProtT5 have, so far in our hands, outperformed those trained on ESM-1b (54) and others (4; 9; 24; 23) for numerous different applications (37-39; 42; 64; 13; 25; 72). This consistency also increases speed as the generation of embeddings has become a limiting step.…”

“…Protein language models (pLMs) are deep learning models pre-trained on large sets of unannotated sequences to generate numerical representations (embeddings) (9; 24; 40; 23; 49; 54). Embeddings from pLMs have been successfully used as input to downstream protein prediction tools (78; 37-39; 41; 42; 44; 64; 14; 25; 27; 63; 72). While some pLM-based methods do not reach the performance of MSA-based methods (23; 38; 72), others exceed those (5; 10; 23; 39; 42; 64; 28; 29; 36).…”