TrendBio

Top 40

View all ›

BACKFILLED Code and industry signals are partial for backfilled days.

Rank 1

ML / AI methods

Benchmarks expose limits of learned models in protein, drug, and enzyme prediction

Recent benchmarks test whether learned molecular models generalize beyond compound memorization, similarity shortcuts, and simpler baselines.

7 preprints · ↑ 2.7× vs 4-wk avg
Rank 2

ML / AI methods

Learned models predict peptide, epitope and protein binding from sequence and structure

Computational screens now span HLA II PTM-peptides, SH2 domains, epitopes and DUF protein pairs, with AlphaFold 3 scaling complex prediction.

8 preprints · ↑ 2.4× vs 4-wk avg
Rank 3

Structural bio & protein design

ML generative and scoring pipelines for antibody, DARPin, and peptide binder design

Blinded prospective benchmarks now test AI-designed antibodies against experimental affinity and developability, moving evaluation beyond in silico scores.

6 preprints · 3 published · ↑ 2.3× vs 4-wk avg
Rank 4

Tools & infrastructure

Software tools now interpret variants across structural, tissue, and isoform contexts

Recent preprints extend variant interpretation to population haplotypes, tissue-specific target genes, transcript isoforms, and AlphaGenome model predictions.

6 preprints · ↑ 2.4× vs 4-wk avg
Rank 5

Genomics

Chromosome-scale genomes reveal fusion, mixing and sex chromosome lability across taxa

Comparative analysis of 5821 chromosome-scale animal genomes and new reference assemblies sharpens views of chromosome evolution and reference-dependent…

4 preprints · 2 published · ↑ 2.1× vs 4-wk avg