Tools & Resources

Sequence-based prediction and analysis of residue-level protein electrostatics and pKa-linked behavior. KaML uses protein language-model representations to support structural interpretation and proteome-scale annotation when experimental structures are unavailable.

Reference. Shen, M., Dayhoff, G. W. II, and Shen, J. “Protein Electrostatic Properties are Fine-Tuned Through Evolution.” Research Square (2026). Under review at Nature Communications.

A multitask sequence-based protein-profiling platform for predicting covalently ligandable cysteines, reversible ligand-binding residues, cysteine functional context, intrinsic disorder, molecular recognition features, and pKa-linked reactivity.

Reference. Dayhoff, G. W., Kortzak, D., Liu, R., Shen, M., Zhang, Z.-Y., and Shen, J. “Illuminating the Ligandable Human Proteome with AI Protein Profiling.” bioRxiv (2026). In peer review at Nature.

An integrated intrinsic-disorder analysis platform combining six established predictors. RIDAO supports individual proteins, comparative studies, and genome-scale workflows while retaining access to predictor-level and consensus results.

Reference. Dayhoff, G. W. II, and Uversky, V. N. “Rapid Prediction and Analysis of Protein Intrinsic Disorder.” Protein Science 31(12), e4496 (2022).

A modular, configuration-driven framework for developing downstream prediction models on pretrained protein language-model backbones. It standardizes dataset construction, supervision refinement, training, evaluation, ensembling, and experiment specification, with explicit attention to information leakage.

A hybrid molecular-dynamics and machine-learning method implemented through a modified GROMACS workflow. The research implementation links TensorFlow through the C API and incorporates learned velocity updates with periodic Hamiltonian integration.

Reference. Dayhoff, G. W. II, and Varma, S. “Velocity Updates from Machine Learning for Stable Molecular Dynamics Integration.” AIP Advances 16, 015205 (2026).

AMBER pmemd24 source modifications supporting the constant-pH and replica-exchange molecular-simulation methods used in this research program.

A FASTA-formatted sequence resource for PDB-centered sequence mapping and downstream bioinformatics workflows.

A text mapping resource connecting gPDB sequence records with UniProt identifiers and related reference workflows.