Scientific tools for biologics research

Predict how short-peptide secondary-structure populations redistribute across water, methanol/water, TFE/water, and SDS micelles from sequence and environment.

Number antibody sequences using IMGT, Kabat, Chothia, or AHo schemes and label framework and CDR residues for sequence review.

Number antibody variable domains, assign V/J germlines, label FR/CDR regions, and locate sequence liabilities.

Design humanized antibody variable domains with Sapiens or CDR grafting, and score exact overlapping 9-mer prevalence against the official OASis human-repertoire database.

Predict 3D structures of antibodies and nanobodies using deep learning for structure-based design.

Compare model-predicted high-concentration viscosity risk for paired antibody heavy and light chains.

Compare two published CDR-only logistic decision scores for heavy-chain nanobody variable domains and retain the exact model, numbering, and biochemical evidence.

Score heavy-chain or paired heavy/light antibody sequences with AbLang2 confidence or pseudo-log-likelihood and compare candidate plausibility.

Map aggregation propensity across a protein structure, identify exposed positive-score regions, and compare residue- and chain-level summaries.

Generate pinned 1,280-dimensional SaProt embeddings and zero-shot mutation preference scores for protein sequences paired with precomputed 3Di structure tokens.

Generate biomolecular binder candidates conditioned on a target structure and design constraints.

Design and score protein sequences against a structure or aligned conformational ensemble with five released inverse-folding model profiles.

Generate protein sequence and all-atom coordinates around an explicit, checksum-bound three-dimensional ligand pose.

Generate target-conditioned protein binders for up to four heterogeneous protein targets in one model-loaded native batch.

Generate all-atom protein, protein-binder, small-molecule-binder, and enzyme-scaffold structures with mode-specific molecular constraints.

Predict target–binder complex cohorts with explicit sampling seeds, per-token PAE, confidence evidence, and checksum-bound structures.

Score predicted binder interfaces from checksum-bound complex structures and matching per-token PAE evidence.

Compare predicted and reference complexes using native DockQ 2.1.3 with explicit chain mapping and alignment policy.

Design protein sequences for complete backbones with ProteinMPNN or condition sequence design on ligands, nucleic acids, metals, and cofactors with LigandMPNN.

Profile antibodies with 24 interpretable sequence, structure, surface, patch, and OASis humanness features, or screen CDR tryptophans with a qualified static SASA/electrostatics heuristic.

Repair checksum-bound PDB structures with PDBFixer and calculate exact collective pairwise RMSD matrices with pyRMSD in one structure-preparation and ensemble-analysis workspace.

Compare antibody sequences against the PLAbDab patent and TheraSAbDab therapeutic reference collections using IMGT-aligned ranked matches and position-level identity.

Generate one 20–50-residue unconditional protein backbone. This tool does not support motif scaffolding, target-conditioned binder design, or symmetric generation.

Design antibody and nanobody structures against a specified antigen using a multistage generative workflow.

Generate unconditional all-atom protein structures with a length, random seed, and direction scale using an equivariant diffusion model.

Predict nanobody structures and review residue-level confidence for VHH candidates.

Measure contact molecular surface and shape complementarity across a protein–protein interface.

Co-design protein sequence and structure around ligands, DNA, or RNA conditioning molecules.

Design protein binders from checksum-bound binder and target templates with a bounded, sequential FoldCraft/ColabDesign trajectory batch.

Iteratively co-optimize protein sequence and structure with the licensed Protenix path, including ligand, DNA, RNA, antibody, symmetry, cyclic, and sequence-constraint modes.

Predict all-atom protein, antibody–antigen, nucleic-acid, ligand, and ion complexes in a bounded sequential CUDA batch.

Generate bounded peptide C-alpha conformational ensembles for checksum-bound MHC-I complexes with the multi-allele MHC-Diff checkpoint.

Predict the published transformed-affinity score and optional unrefined peptide–MHC-I structure for HLA-A*02:01 9-mers.

Predict HLA-II presentation probabilities and allele-specific 9-mer core scores with the published two-fold ensemble.

Score protein sequences with the global-population DRB1 pIRS model and optional exact human-reference 9-mer filtering.

Predict MHC-I binding affinity, antigen-processing, and presentation scores across a bounded peptide and allele panel.

Predict sequence-level aggregation propensity and residue-level candidate aggregation-prone regions with the five-fold PALM ensemble.

Compute mean-pooled protein embeddings with the published ESMC 300M or 600M checkpoints on the shared CPU runtime.

Run large ESM embeddings, masked protein generation, and ESMFold2 structure prediction.

Score antibody variable regions and rank bounded framework substitutions with exact 8–12-mer human and mouse reference libraries from a frozen current-OAS snapshot.