Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
Rank 1
Single-cell
Models predict gene expression and drug responses from perturbation and tissue data
Single-cell perturbation atlases now reach 100M transcriptomes, providing training substrates for predictive models of drug and gene effects.
4 preprints · 3 published · ↑ 2.7× vs 4-wk avg
Rank 2 Single-cell
Multimodal models couple single-cell transcriptomes with regulatory and sequence priors
Deep learning frameworks now link single-cell transcriptomes with TCR, sequence, chromatin, and regulatory information across modalities.
8 preprints · 3 published · ↑ 2.0× vs 4-wk avg
Rank 3 ML / AI methods
Estimators that model mixed-type and partially observed biomedical tables directly
Biobank-scale omics and clinical tables with mixed types and missing values are driving estimators that encode their structure directly.
6 preprints · 1 published · ↑ 1.9× vs 4-wk avg
Rank 4 Tools & infrastructure
Enrichment and overlap testing increasingly organize genomic interval analyses
Recent preprints benchmark interval-based peak calling and overlap statistics, and extend alignment-based transcript quantification.
4 preprints · 2 published · ↑ 1.9× vs 4-wk avg
Rank 5 ML / AI methods
Neural foundation and generative models tested for transfer and input attribution
Recent preprints test whether large neural models transfer across animals and whether their input effects hold under matched controls.
4 preprints · ↑ 1.9× vs 4-wk avg
Rank 6 Deep learning for molecular property prediction with limited or shifted data↑ 1.6×ML / AI methodsRank 7 Structural and learned protein context expands viral dark matter detection↑ 1.5×Structural bio & protein designRank 8 Automated pipelines and precomputed resources scale microbial and fungal genome analysisTools & infrastructureRank 9 Modeling proteoform detectability and uncertainty beyond fixed database assumptionsProteomicsRank 10 Multimodal models fuse histology, genomics, and protein data for biological predictionML / AI methodsRank 11 Controlling pipeline-level bias in omics and spectral inference methodsML / AI methodsRank 12 Connectivity constrains neural functional states and effective-circuit inferenceSystems biologyRank 13 Biomedical LLM systems assessed for evidence grounding rather than endpoint answersML / AI methodsRank 14 Haplotype reconstruction from long and short reads for variant detectionGenomicsRank 15 Explicit model assumptions shape stochastic evolutionary and population simulationsSystems biologyRank 16 Protein language models are steered by experimental rewards and sampling strategyStructural bio & protein designRank 17 Structured constraints on single-cell expression and regulatory inference estimatesSingle-cellRank 18 Benchmarks test pipelines that build and refine cryo-EM and protein structural modelsStructural bio & protein designRank 19 Generative models conditioned on protein structure and fragments drive ligand designStructural bio & protein designRank 20 Genetic background and aging jointly shape DNA methylation variation across scalesEpigenomics
Show 14 more
Rank 21 Validity criteria for compressed and explained cell microscopy modelsML / AI methodsRank 22 Graph-based inference of molecular interactions across omics and regulatory layersSystems biologyRank 23 Conditional generative models extend to RNA, imaging, and conformational ensemblesML / AI methodsRank 24 Reduced-resolution models link nucleic acid sequence to condensate and phase behaviorStructural bio & protein designRank 25 Structure-aware models rank non-coding and sub-genic variant tolerance across the genomeGenomicsRank 26 Docking, deep learning and MD resolve inhibitor binding at NDM-1, ABCG2 and tubulinStructural bio & protein designRank 27 Residue-resolved interactome mapping and cell-cycle proteome deconvolutionProteomicsRank 28 Databases and algorithms make non-canonical RNA and DNA structure features searchableTools & infrastructureRank 29 Graph learning over protein and drug networks for target prioritisation and repurposingSystems biologyRank 30 Throughput-focused computation for tumor WGS, dosimetry reads, and radiomicsGenomicsRank 31 Learned and physics-based models drive iterative search in strain and enzyme designStructural bio & protein designRank 32 Computational models correct spatial transcriptomics and infer tissue domains and stressTranscriptomicsRank 33 Learned continuous descriptors link cell and network structure to biological statesSystems biologyRank 34 Interpretable models tie peptide-HLA binding to learned energetics and structureStructural bio & protein design