Top 40
View all ›BACKFILLED Code and industry signals are partial for backfilled days.
Rank 1
Single-cell
Generative and graph models infer cell states and gene structure from omics data
New generative and graph-based methods integrate unpaired omics and benchmark single-cell and spatial inference tasks in late 2025.
14 preprints · 13 published · ↑ 3.9× vs 4-wk avg
Rank 2 ML / AI methods
Probing internal representations of biological models to steer design and diagnose bias
Biological foundation models are now being inspected via sparse autoencoders, layer probing, and explanation methods, exposing confounders and motif bias.
9 preprints · 2 published · ↑ 2.4× vs 4-wk avg
Rank 3 ML / AI methods
Auditing genomic deep learning models for confounders, motif bias and leakage
Genomic and protein deep learning models are now being evaluated for what they learn, not only for headline accuracy.
8 preprints · 2 published · ↑ 2.3× vs 4-wk avg
Rank 4 Genomics
Callable-loci and repeat-aware tools reformulate population-genetic and variant analyses
Recent tools extend variant calling, tandem repeat genotyping, and selection inference to callable-site and repeat-aware formulations.
4 preprints · 5 published · ↑ 2.2× vs 4-wk avg
Rank 5 ML / AI methods
Taxonomic tree priors and compositional models refine microbiome data analysis
Recent preprints and a published method pair prior biological structure with microbiome modeling for differential abundance, augmentation, and prediction.
5 preprints · 1 published · ↑ 1.9× vs 4-wk avg
Rank 6 Pretrained embeddings drive deep models of enzyme kinetics and ligand binding↑ 2.0×ML / AI methodsRank 7 Mechanical loading and stress inference coupled to multiscale tissue models↑ 2.0×Systems biologyRank 8 Parallel and out-of-core tools accelerate genomic and single-cell data processing↑ 1.9×Tools & infrastructureRank 9 Spatial multi-omics resolves prognostic tumor niches and cell-state heterogeneity↑ 1.9×TranscriptomicsRank 10 Benchmarking and adapting omics methods for depth, batch, and sequencing-error limits↑ 1.9×GenomicsRank 11 Fitted brain and language models are scored against independent neural recordings↑ 1.9×Systems biologyRank 12 Integrated omics frameworks prioritize biomarkers and candidate disease genes↑ 1.7×Systems biologyRank 13 Reduced dynamical models of cell size, actin networks, and gene expression bursting↑ 1.8×Systems biologyRank 14 Computational methods recover protein information from mixed or compositional data↑ 1.7×ProteomicsRank 15 Parameter inference and identifiability in mechanistic biological models↑ 1.7×Systems biologyRank 16 Calibrated fine-mapping and regulatory variant prioritization refine GWAS inference↑ 1.7×GenomicsRank 17 Benchmarked calibration of structural tools for interfaces, flexibility, and foldingStructural bio & protein designRank 18 Sequence classification and search tools benchmarked for speed, accuracy and compressionTools & infrastructureRank 19 Graph and language models encode network context to predict omics perturbation responsesSingle-cellRank 20 Computational methods resolve parental, chromosomal and fetal genetic contributions↑ 1.5×Genomics
Show 20 moreShow less
Rank 21 Protein language model embeddings reveal structure, function, and homology from sequenceML / AI methodsRank 22 Quantifying topology, calibration, and model uncertainty in phylogenetic inferenceSystems biologyRank 23 Software frameworks for isotope flux, microbial interaction and microbiome analysisMetabolomicsRank 24 Language models predict trial, cancer and chronic disease outcomes from biomedical dataML / AI methodsRank 25 Open-source tools for population genetics, pangenome graphs, and sequence classificationTools & infrastructureRank 26 Linkage, biophysical stability and weak selection shape mutation fitness inferenceGenomicsRank 27 Chromosome-scale assemblies resolve introgression and hybrid genome historyGenomicsRank 28 Tracking individual cells and lineages across time-lapse imaging and atlas dataSingle-cellRank 29 Generative models steered toward target properties for peptides, codons, and enzymesStructural bio & protein designRank 30 LLMs Benchmarked for Microbial Traits, Variant Triage and Pipeline GenerationML / AI methodsRank 31 Network and contrastive methods infer microbe-drug, drug-drug and multi-omics linksSystems biologyRank 32 Deep learning models link spatial, omics, and imaging profiles to cancer outcomesML / AI methodsRank 33 Non-model animal genome assemblies are being paired with gene-level annotationGenomicsRank 34 Deep learning extracts molecular structure from single-molecule and cryo-EM imagesStructural bio & protein designRank 35 Gene circuit control via optogenetics, pattern switching and neuromorphic designSystems biologyRank 36 Diffusion models sample protein conformations and probe learned energeticsStructural bio & protein designRank 37 Scalable graph, haplotype and lineage inference across large genomic datasetsGenomicsRank 38 Lipids, ligands and glycans reshape protein conformational energetics and dynamicsStructural bio & protein designRank 39 Generative and optimization-based RNA design now targets 3D and pseudoknotted foldsStructural bio & protein designRank 40 AI frameworks extend de novo design to antibodies, proteins, and small-molecule hitsStructural bio & protein design