Research map / Probabilistic and scientific modeling
Molecules and proteins: research map
1,990 accepted papers on Molecules and proteins in Probabilistic and scientific modeling, from ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026), grouped into 6 clusters and 26 approaches. The busiest year so far is 2026.
Within Probabilistic and scientific modeling, its share grew from 24.1% in 2023–24 to 30.4% in 2025–26 (490 → 927 papers at ICML, NeurIPS, CVPR and AAAI, the venues with data for all four years).
Explore Molecules and proteins in the interactive map
Working on something in this topic? Describe your idea in scime atlas to see which approach it falls under, the closest papers by meaning and how crowded the spot has become.
Approaches and key papers
decision · machine · statistical · 586 papers
Approaches in this cluster:
- Prediction-powered and conformal inference (129 papers)
Gives valid statistical inference using conformal, prediction-powered and multiple-testing methods. - Statistical analysis of ML practice (122 papers)
Applies survival analysis, additive models and variable selection to understand reproducibility and importance. - Machine learning for physical simulation (96 papers)
Builds datasets, benchmarks and graph models for fluid and multiphysics simulation and design. - Counterfactual and statistical explanations (81 papers)
Generates counterfactual and logic-based explanations and frames explainability as inference. - Sparse concepts and hypothesis testing (78 papers)
Defines and tests emergent sparse interactions and statistical significance in networks. - Cold-start recommendation models (80 papers)
Handles new users and items using dropout and Bayesian factorization models.
Most cited and most cited since 2024:
- ClimaX: A foundation model for weather and climate (ICML 2023 · 167 citations)
- DropoutNet: Addressing Cold Start in Recommender Systems (NeurIPS 2017 · 149 citations)
- PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (NeurIPS 2025 · 18 citations)
- Adversarial Socialbots Modeling Based on Structural Information Principles (AAAI 2024 · 15 citations)
latent · space · representations · 441 papers
Approaches in this cluster:
- Neuro-symbolic probabilistic layers (120 papers)
Combines probabilistic structured prediction and exact inference with neural networks. - Sparse graphical and tensor models (97 papers)
Learns sparse graphs and tensor decompositions for high-order interaction data. - Probabilistic embeddings and relational inference (93 papers)
Uses box embeddings, analogical embeddings and lifted inference for relational data. - Equivariant and disentangled latent representations (90 papers)
Learns group-structured, equivariant or disentangled latent spaces via symmetry priors. - Generative latent models for single-cell data (41 papers)
Models cellular perturbations and hierarchies with VAEs and concept bottlenecks.
Most cited and most cited since 2024:
- Complex Embeddings for Simple Link Prediction (ICML 2016 · 1,122 citations)
- Discriminative Embeddings of Latent Variable Models for Structured Data (ICML 2016 · 499 citations)
- Neural Markov Random Field for Stereo Matching (CVPR 2024 · 44 citations)
- SketchINR: A First Look into Sketches as Implicit Neural Representations (CVPR 2024 · 11 citations)
proteins · binding · sequence · 359 papers
Approaches in this cluster:
- Generative protein structure design (142 papers)
Generates protein structures and sequences using flow matching, equivariant models and AlphaFold. - Protein fitness and variant effect prediction (71 papers)
Predicts mutation effects using language models and autoregressive transformers. - Protein interface and docking prediction (72 papers)
Predicts interfaces, ligand docking and conformations with equivariant 3D structure networks. - Protein representation learning (40 papers)
Learn protein representations from sequence and 3D structure graphs. - Biological sequence transformers (34 papers)
Applies transformer language models to protein and genomic sequences.
Most cited and most cited since 2024:
- Language models enable zero-shot prediction of the effects of mutations on protein function (NeurIPS 2021 · 699 citations)
- TANKBind: Trigonometry-Aware Neural NetworKs for Drug-Protein Binding Structure Prediction (NeurIPS 2022 · 171 citations)
- AlphaFold Meets Flow Matching for Generating Protein Ensembles (ICML 2024 · 88 citations)
- SE(3)-Stochastic Flow Matching for Protein Backbone Generation (ICLR 2024 · 15 citations)
molecules · drug · chemical · 306 papers
Approaches in this cluster:
- Structure-based 3D molecule generation (107 papers)
Generates drug-like molecules in protein pockets and extends scaffolds with 3D generative models. - Diffusion for molecular conformations (67 papers)
Generates 3D conformers using gradient fields, geometric diffusion and neural dynamics. - Machine-learned interatomic potentials (74 papers)
Learns and benchmarks interatomic potentials for atomistic simulation of materials. - Graph models for retrosynthesis (10 papers)
Predict retrosynthesis using graph-based generation and attention. - Structure-based molecule design with diffusion (48 papers)
Generate pocket-conditioned and synthesizable ligands with diffusion on 3D structure.
Most cited and most cited since 2024:
- SchNet: A continuous-filter convolutional neural network for modeling quantum interactions (NeurIPS 2017 · 470 citations)
- DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking (ICLR 2023 · 362 citations)
- Learning Smooth and Expressive Interatomic Potentials for Physical Property Prediction (ICML 2025 · 33 citations)
- GxVAEs: Two Joint VAEs Generate Hit Molecules from Gene Expression Profiles (AAAI 2024 · 12 citations)
flow · trajectory · matching · 222 papers
Approaches in this cluster:
- Flow matching generative modeling (97 papers)
Develops flow matching theory and variants, including equivariant and spectral versions. - Multimodal trajectory prediction (95 papers)
Predicts agent trajectories using causal disentanglement, trajectory sets and graph models. - Spatiotemporal forecasting and imputation (30 papers)
Forecasts or imputes spatiotemporal data with graph-informed flow matching and residual diffusion.
Most cited and most cited since 2024:
- FlowNet 2.0: Evolution of Optical Flow Estimation With Deep Networks (CVPR 2017 · 3,371 citations)
- Social GAN: Socially Acceptable Trajectories With Generative Adversarial Networks (CVPR 2018 · 2,462 citations)
- DiffCast: A Unified Framework via Residual Diffusion for Precipitation Nowcasting (CVPR 2024 · 68 citations)
- BAT: Behavior-Aware Human-Like Trajectory Prediction for Autonomous Driving (AAAI 2024 · 64 citations)
event · point processes · temporal · 76 papers
Approaches in this cluster:
- Neural point-process models (64 papers)
Models event sequences with attention, latent processes and Hawkes or graphical event models. - Survival analysis under censoring (12 papers)
Models time-to-event data with copulas, adversarial and likelihood methods handling censoring.
Most cited and most cited since 2024:
- The Neural Hawkes Process: A Neurally Self-Modulating Multivariate Point Process (NeurIPS 2017 · 204 citations)
- Event Probability Mask (EPM) and Event Denoising Convolutional Neural Network (EDnCNN) for Neuromorphic Cameras (CVPR 2020 · 97 citations)
- EBS-EKF: Accurate and High Frequency Event-based Star Tracking (CVPR 2025 · 7 citations)
- Deep Copula-Based Survival Analysis for Dependent Censoring with Identifiability Guarantees (AAAI 2024 · 7 citations)
Related topics in Probabilistic and scientific modeling
- Dynamical systems (969)
- Uncertainty and Gaussian processes (1,162)
- Variational and generative sampling (2,060)
- Treatment effects (421)
- Differential equations and PDEs (614)
- Causal discovery (830)
- Bayesian optimization (397)