Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
Rank 1
Structural bio & protein design
Guided generative models steer protein and molecule design with structural feedback
Recent preprints and publications apply guiding potentials and iterative feedback to protein and molecule generative design.
8 preprints · 2 published · ↑ 2.7× vs 4-wk avg
Rank 2 Single-cell
Inferring regulatory programs and cell states from single-cell and spatial data
Single-cell ATAC, RNA, and spatial datasets now support scalable inference of regulatory activity, pathway, and cell-type communication programs.
9 preprints · 4 published · ↑ 2.7× vs 4-wk avg
Rank 3 ML / AI methods
Reducing patch volume and annotation needs in histopathology analysis
Recent histopathology work seeks to cut patch volume, annotation burden, or input scale while targeting tumour segmentation and prediction tasks.
12 preprints · 2 published · ↑ 2.4× vs 4-wk avg
Rank 4 ML / AI methods
Protein language model embeddings fused with structure, binding sites, and chemistry
Pretrained protein language model embeddings are increasingly combined with 3D structure, binding-site, text, and chemical encoders in shared latent spaces.
6 preprints · 1 published · ↑ 1.9× vs 4-wk avg
Rank 5 Structural bio & protein design
Benchmarking and automating structure-based modeling of biomolecular complexes
Docking, MD, and generative design tools are being benchmarked and packaged into automated workflows for antibodies, PROTACs, and RNA targets.
4 preprints · 3 published · ↑ 1.9× vs 4-wk avg
Rank 6 Benchmarks test small LLM agents and new Python tools for transcriptomic analysis↑ 1.9×Tools & infrastructureRank 7 Chromatin readouts map regulatory programs and 3D genome organization across cell types↑ 1.8×EpigenomicsRank 8 Latent models infer single-cell dynamics, gene regulation, and RNA-protein kinetics↑ 1.9×Single-cellRank 9 Computational scrutiny of RNA-chromatin artifacts and intronic polyadenylation sites↑ 1.8×EpigenomicsRank 10 Standardized tools make proteomics and omics analysis reproducible and traceable↑ 1.8×Tools & infrastructureRank 11 Pretrained language model embeddings feed RNA and miRNA prediction models↑ 1.7×ML / AI methodsRank 12 Computational models refine cell typing, marker detection and spatial RNA localization↑ 1.6×Single-cellRank 13 Learned and model-free methods map RNA modifications and non-coding variant effects↑ 1.6×GenomicsRank 14 Proteome-wide models predict interactions, disorder grammars and segment boundariesML / AI methodsRank 15 Multimodal and interpretable models predict compound toxicity and bioactivity endpointsML / AI methodsRank 16 Trajectory constraints improve regulatory dynamics inference from single-cell dataSingle-cellRank 17 Generative and language models now design protein, peptide, and mRNA sequencesML / AI methodsRank 18 Marker genes guide phylogeny, metagenome coassembly, and genome quality assessmentGenomicsRank 19 Learned models operate directly on raw microscopy and cryo-EM observationsML / AI methodsRank 20 Analysis tools for DIA proteomics, MS metadata, and metabolite annotationMetabolomics
Show 20 moreShow less
Rank 21 Spatial neighborhood context enters perturbation modeling and differential expressionTranscriptomicsRank 22 Multimodal LLMs couple omics, DNA, and protein data with natural-language queryingML / AI methodsRank 23 Knowledge graphs underpin deployable biomedical prediction and literature retrieval toolsTools & infrastructureRank 24 MD resolves loop flexibility and allosteric coupling across viral and bacterial proteinsStructural bio & protein designRank 25 Alignment-light methods advance prokaryotic genome and metagenomic analysisGenomicsRank 26 Computational models track pathogen genome evolution, recombination and transmissionGenomicsRank 27 Genome mining reveals antimicrobial peptides and effectors from microbial genomesGenomicsRank 28 Learned models link transcription, annotation bias and developmental gene regulationTranscriptomicsRank 29 Long-read tools add methylation linkage, standardized storage and benchmarking to genome…GenomicsRank 30 ML models map sequence and structure to ligand, pocket, and peptide-MHC recognitionStructural bio & protein designRank 31 Computational and proteomic profiling resolve metabolic pathway structure and fluxSystems biologyRank 32 Learned representations probe biomolecular kinetics and metastable states from simulationStructural bio & protein designRank 33 Finite element and surrogate models couple microstructure to tissue mechanical outputSystems biologyRank 34 Computational models link EEG dynamics to sleep, consciousness, and entrainmentSystems biologyRank 35 Experimental benchmarks test folding energy and sequence models for protein variantsStructural bio & protein designRank 36 Persistent homology and Grassmann geometry recover multiscale structure in omics dataSingle-cellRank 37 Chromosome-scale nutcracker genome and spruce SNP array for population geneticsGenomicsRank 38 Viscoelastic aging and sticker-spacer modeling define condensate material statesStructural bio & protein designRank 39 Emergent system-level constraints set survival and stability outcomes across modelsSystems biologyRank 40 Brain imaging representations engineered for subject, patient, or topological propertiesSystems biology