Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
Rank 1
Epigenomics
Sequence-based models link DNA motifs, 3D chromatin structure and accessibility
Recent preprints and papers pair sequence foundation models with Hi-C and nucleosome data to predict chromatin accessibility, structure and methylation.
6 preprints · 4 published · ↑ 4.3× vs 4-wk avg
Rank 2 Genomics
Mobile element inventories gaining benchmarks and ecological host context
Soil plasmid resources and virus-host Hi-C benchmarks now tie mobile element inventories to host and ecological context.
4 preprints · 5 published · ↑ 3.2× vs 4-wk avg
Rank 3 Single-cell
Computational methods denoise, impute and map single-cell transcriptomic data
Recent work pairs deep learning and statistical models with scRNA-seq, FISH and single-cell chromatin data to denoise, assign and map cells.
3 preprints · 6 published · ↑ 2.3× vs 4-wk avg
Rank 4 ML / AI methods
Learned sequence models generate context-conditioned protein designs
Recent preprints and publications apply protein language models and LLMs to protein design, secretion engineering, and allergenicity prediction.
4 preprints · 2 published · ↑ 2.3× vs 4-wk avg
Rank 5 Genomics
Learned models score noncoding regulatory elements, variants, and transcripts
Recent reports apply trained sequence models and calibrated scores to prioritize noncoding variants, regulators, and transcript isoforms.
1 preprint · 4 published · ↑ 2.1× vs 4-wk avg
Rank 6 Multimodal graph neural networks predict compound properties and bioactivity↑ 2.0×ML / AI methodsRank 7 Learned protein embeddings predict interactions, affinity and structure reliability↑ 1.9×Structural bio & protein designRank 8 Sampling frames shape which pathogen lineages enter epidemiological inference↑ 1.9×Tools & infrastructureRank 9 Model form and data fusion shape clinical risk and dosing predictions↑ 1.9×ML / AI methodsRank 10 Transformer and contrastive models predict RNA, protein and peptide interaction sites↑ 1.8×ML / AI methodsRank 11 Targeted long-read assembly and matched references sharpen genome-scale analysis↑ 1.8×GenomicsRank 12 Deep learning models are trained on and benchmarked against genomic variation↑ 1.7×GenomicsRank 13 Open-source pipelines package sequencing and omics analyses into reproducible workflows↑ 1.7×Tools & infrastructureRank 14 Open tools link simulation, prediction, and design for protein flexibility and binding↑ 1.6×Structural bio & protein designRank 15 Aligning unpaired and spatial omics modalities via optimal transport and graph geometrySingle-cellRank 16 Omitted epistasis inflates GWAS nulls; noise QTLs and metQTLs map complex-trait loci↑ 1.6×GenomicsRank 17 Stochastic and geometric local rules shape cell migration and membrane protein transport↑ 1.5×Systems biologyRank 18 Learned brain graph models decode EEG, SEEG, fMRI, and structural connectome data↑ 1.6×ML / AI methodsRank 19 Bayesian and kernel methods infer interaction and noise structure from biological data↑ 1.6×ML / AI methodsRank 20 Metabolic models integrate multi-omics to infer disease-specific metabolic statesSystems biology
Show 20 moreShow less
Rank 21 Benchmarked computational workflows for kinase, interaction and proteoform mappingProteomicsRank 22 Structure-constrained learned encoders predict unmeasured genes and perturbation responsesML / AI methodsRank 23 Standalone tools speed up sequence search, read simulation and metagenomic analysisTools & infrastructureRank 24 Causal and network methods prioritize genes, modules and disease driversSystems biologyRank 25 Computational methods infer cell identity and dynamics from single-cell and spatial dataSingle-cellRank 26 Ligand binding and conformational dynamics in disordered and allosteric proteinsStructural bio & protein designRank 27 Minimizer density modeling and compressed indexing for scalable DNA sequence analysisTools & infrastructureRank 28 Benchmarks and ML models resolve paired immune receptor chain pairing and specificityML / AI methodsRank 29 Excitability and state transitions framed as phase-plane and maximum entropy dynamicsSystems biologyRank 30 Inferring dynamic regulatory networks and virtual cell states from single-cell dataSingle-cellRank 31 Model-based inference of reaction network dynamics from noisy and time-course dataSystems biologyRank 32 Allosteric Networks Govern Ligand Recognition in Laccase and p53 Y220CStructural bio & protein designRank 33 Learned protein and sequence representations applied to viral and eukaryotic genomesGenomicsRank 34 Network and competing-history models infer microbial evolutionary structureSystems biologyRank 35 Target-conditioned generative models produce ligands, peptides, and cell morphologiesStructural bio & protein designRank 36 Parameterized network models fit learning rules and circuit architecture to neural dataSystems biologyRank 37 Genome-wide inference of duplication history, chromosome fusion, and hybrid couplingGenomicsRank 38 New tools visualize RNA pseudoknots, recombination graphs, and gene loss in genomesTools & infrastructureRank 39 Generative and hierarchical models align multimodal mouse brain data to reference atlasesTools & infrastructureRank 40 Generative design of peptide and ligand binders conditioned on 3D target structureStructural bio & protein design