| Structure Prediction |
| Deep learning-based protein structure prediction | High-accuracy 3D folding from sequence data; catalytic site inference | | https://www.deepmind.com/research/highlighted-research/alphafold (accessed on 1 October 2025) |
| Multi-track neural network integrating sequence and structural information | Predicts structure and function; supports novel fold design | | https://boinc.bakerlab.org/rosetta/ (accessed on 1 October 2025) |
| Enhanced structure prediction using multimodal inputs (MSAs, templates, embeddings) | Robust prediction across diverse biomolecules | | https://neurosnap.ai/service/Chai-1 (accessed on 1 October 2025) |
| Structure prediction without multiple sequence alignments (MSAs) | Accurate, MSA-free folding for low-homology sequences | | https://github.com/HeliXonProtein/OmegaFold orhttps://cosmic-cryoem.org/tools/omegafold/ (accessed on 1 October 2025) |
| Fast, efficient structure prediction framework | Optimised for industrial biodesign pipelines | | https://github.com/PaddlePaddle/PaddleHelix/tree/dev/apps/protein_folding/helixfold (accessed on 26 October 2025) |
| Speed-optimised AlphaFold implementation | GPU-parallelisation for rapid inference | | https://github.com/hpcaitech/FastFold (accessed on 1 October 2025) |
| Protein and Enzyme Design |
| AI-guided enzyme design platform | Generates and screens enzyme variants with improved catalytic traits | | https://neurosnap.ai/service/NeuroFold (accessed on 1 October 2025) |
| Transformer-based language model for protein generation | De novo protein sequence generation preserving function | | https://github.com/salesforce/progen (accessed on 1 October 2025) |
| Large-scale protein language model | Sequence embeddings, mutation effect, and function prediction | | https://github.com/facebookresearch/esm (accessed on 1 October 2025) |
| Diffusion-based generative model for protein backbones | Enables creation of novel folds and catalytic sites | | https://github.com/RosettaCommons/RFdiffusion(accessed on 1 October 2025) |
| Sequence design conditioned on backbone structures | Rapid backbone-to-sequence mapping for design tasks | | https://github.com/dauparas/ProteinMPNN (accessed on 1 October 2025) |
| Controlled generative framework for structural design | Conditional structure generation with user-specified features | | https://generatebiomedicines.com/chroma (accessed on 26 October 2025) |
| Catalytic Site, Substrate Specificity and Metal-Binding Prediction |
| | Predicts catalytic metal ions in enzymes | | https://mahomes.ku.edu/help (accessed on 1 October 2025) |
| | Predicts functional and substrate classes of adenylate-forming enzymes | | https://github.com/serina-robinson/adenylpred (accessed on 1 October 2025) |
| | Predicts turnover, stability, and stereoselectivity | PEACCEL (The AI company for Life Science) | https://www.peaccel.com/technology/innovsar-artificial-intelligence-platform/ (accessed on 1 October 2025) |
| | 3D active-site and sequence generation | | https://github.com/Shen-Lab/gcWGAN (accessed on 1 October 2025) |
| Molecular Docking |
| Diffusion-based molecular docking model | Flexible ligand–protein docking; 3D pose generation | | https://github.com/gcorso/DiffDock (accessed on 1 October 2025) |
| Convolutional neural network (CNN)-based docking framework | ML-enhanced scoring and ligand ranking | | https://github.com/gnina/gnina or https://proteiniq.io/app/gnina (accessed on 26 October 2025) |
| AI-based pocket prediction and docking | Predicts binding pockets and supports flexible docking | | https://proteiniq.io/app/pocketflow or https://github.com/Saoge123/PocketFlow (accessed on 26 October 2025) |
| Classical open-source docking program | Widely used for small-molecule and enzyme–ligand interactions | | http://vina.scripps.edu (accessed on 26 October 2025) |
| In silico Mutagenesis |
| Predicts the functional impact of mutations | High-throughput mutational scanning via deep learning | | https://github.com/gamazonlab/DeepMutScan (accessed on 26 October 2025) |
| Predicts protein–protein interaction changes upon mutation | Estimates ΔΔG and interface disruption | | https://github.com/guangyu-zhou/MuPIPR (accessed on 26 October 2025) |
| Inverse folding model (structure → sequence) | Recovers sequences from 3D backbones | | https://neurosnap.ai/service/ESM-IF1 (accessed on 1 October 2025) |
| Predicts mutation effects on stability and dynamics | Visualises conformational and flexibility changes | | https://biosig.lab.uq.edu.au/dynamut/ (accessed on 26 October 2025) |
| Sequence and Structure Analysis |
| Protein language models for representation learning | Embeddings for annotation, classification, and alignment | ProtNLM by Google Research/EBI; ProtBERT by Brandes et al.; ESM-1b by Meta AI/Hugging Face | https://310.ai/docs/function/protnlm (accessed on 1 October 2025); https://huggingface.co/Rostlab/prot_bert (accessed on 26 October 2025); https://huggingface.co/facebook/esm1b_t33_650M_UR50S (accessed on 26 October 2025) |
| Protein sequence clustering and alignment | Scalable similarity search and redundancy reduction | MPI for Developmental Biology | https://toolkit.tuebingen.mpg.de/tools/mmseqs2 (accessed on 1 October 2025) |
| Structure-based search engine | Fast comparison of protein 3D structures | | https://search.foldseek.com/search (accessed on 1 October 2025) |
| Multiple sequence alignment algorithm | High-speed, accurate alignment for large datasets | | https://mafft.cbrc.jp/alignment/software/ (accessed on 1 October 2025) |
| Hidden Markov Model-based alignment and domain search | Detects conserved motifs and functional domains | | http://hmmer.org (accessed on 1 October 2025) |
| Enzymatic Classification and Functional Prediction |
| CNN-based enzymatic classifier | Predicts complete EC numbers from sequence | | https://bitbucket.org/kaistsystemsbiology/deepec(accessed on 1 October 2025) |
| | Predicts full/partial EC numbers | | https://ecpred.kansil.org/ (accessed on 1 October 2025) |
| Expression and Safety Tools |
| Predicts protein solubility using deep learning | AI-based solubility and expression classifier | | https://services.healthtech.dtu.dk/service.php?NetSolP-1.0(accessed on 1 October 2025) |
| Predicts heterologous expression efficiency | Assists codon optimisation and solubility enhancement | | https://tisigner.com/sodope (accessed on 1 October 2025) |
| ToxinPred2 (v2.0)/ADMET-AI (v1.4.0)/eTox (v0.97) | Predicts toxicity, allergenicity, and pharmacokinetics | Comprehensive in silico safety and ADMET profiling | | https://webs.iiitd.edu.in/raghava/toxinpred2/ (accessed on 1 October 2025) |
| Antibody and Binder Design |
| AI-based design of antibodies, peptides, and nanobodies | Generates high-affinity binders and optimises stability | | https://neurosnap.ai/service/NeuroBind (accessed on 1 October 2025) |
| Deep-learning tool for antibody CDR loop modelling | Rapid and accurate loop conformation prediction | Oxford Protein Informatics Group | https://github.com/oxpig/ABlooper (accessed on 1 October 2025) |
| Transformer model for antibody structure prediction | Sequence-to-structure mapping for immunoglobulins | | https://github.com/Graylab/IgFold (accessed on 26 October 2025) |
| Lightweight nanobody structure predictor | Fast prediction of VHH domains and single-chain binders | | https://github.com/dina-lab3D/NanoNet (accessed on 26 October 2025) |
夜雨聆风
