2026-01-01

Top 40

BACKFILLED Code and industry signals are partial for backfilled days.

Single-cell

Learned models infer cell-state dynamics and tissue architecture from single-cell data

Recent preprints and publications apply deep learning to single-cell and spatial data at scale, with OmniCell pretrained on 67M profiles.

Genomics

Multilocus models sharpen heritability, rare-variant and cell-type-resolved inference

Recent GWAS, TWAS, and burden studies extend polygenic and multilocus models across ancestries, cell types, and noncoding rare variants.

Structural bio & protein design

Pretrained structure models repurposed for de novo protein and peptide binder design

Late-December 2025 preprints pair pretrained all-atom predictors and reinforcement or distillation objectives with de novo binder and pocket design.

Single-cell

Deep learning recovers latent temporal structure from static single-cell profiles

Static, untimed, and sparsely sampled single-cell profiles increasingly require learned models to recover phase, context, and spatial structure.

Genomics

Mining genome collections to link microbial sequences with defined systems and traits

Millions of bacterial genomes and metagenomes now support tools that annotate systems, score graph representativeness and profile resistance genes.

6 Theory links epigenetic marks to chromatin mechanics and single-cell state transitionsEpigenomics 7 Deep learning models predict drug, epitope, peptide and molecular propertiesML / AI methods 8 Heterogeneous graph attention and path models predict drug and gene interactionsML / AI methods 9 Generative and ensemble models are prioritising antimicrobial candidates in silicoML / AI methods 10 Learned and ensemble models extend protein sequence, structure and interaction predictionTools & infrastructure 11 Transformer and foundation models are learning microbial traits from genomes and…ML / AI methods 12 Generative and causal models predict and perturb regulatory genomic profilesML / AI methods 13 Enhanced sampling and free energy methods resolve protein conformational determinantsStructural bio & protein design 14 Statistical and Bayesian tools for inferring biology from omics and mutation dataTools & infrastructure 15 Machine learning models classify disease state from omics and brain connectivity dataML / AI methods 16 Mechanistic models of collective cell migration and tissue-scale transportSystems biology 17 Variant-aware CRISPR guide design and conserved Cas13b guides for Dengue virusML / AI methods 18 Geometry-aware inverse folding and blind benchmarks advance RNA structure modelingStructural bio & protein design 19 Packaged pipelines for cell-free chromatin, survival-marker and RNA-seq analysisTools & infrastructure 20 Haplotype-aware tools and assemblies refine Y-chromosome, CNV, and genome reconstructionGenomics

Show 8 more

21 Benchmarks test AI co-folding, RNA structure and antibody developability predictionsStructural bio & protein design 22 Learned models now drive cryo-EM density refinement and electron microscopy acquisitionStructural bio & protein design 23 Upstream feature curation and priors shape single-cell annotation and embedding accuracyML / AI methods 24 Explicit transport and microenvironment constraints shape in silico therapy responseSystems biology 25 Cell-type-resolved atlases link expression variation and aging splicing across speciesSingle-cell 26 Reproducible, provenance-tracked software packaging for curated biological data analysisTools & infrastructure 27 Targeted enrichment tools improve primer design and viral and amplicon detectionGenomics 28 Imposing explicit statistical structure on high-dimensional molecular measurementsSystems biology