Quartz 4

❯

❯

❯

Datasets

Created Sep 26, 2026Modified Sep 07, 2026

Dataset resources, composition, clustering, taxonomic bias, negative examples, synthetic data, and coverage.

52 items with this tag.

Aug 25, 2026
Structural tokens improve zero-shot variant effect prediction in ESM3, but only when structures are not computationally derived
Aug 25, 2026
Structural datasets overrepresent a small number of epitopes
- evidence/datasets
- antibodies/nanobodies
Aug 19, 2026
Pretraining contributes nearly nothing to performance when fine-tuning protein language models in data-rich situations
- training/fine-tuning
- evidence/datasets
Aug 19, 2026
Training inverse folding and diffusion models exclusively on predicted protein structures worsens performance due to how locally perfect they are
Jul 28, 2026
Inverse folding models favor archaeal proteins while sequence-only PLMs favor eukaryotic proteins
- evidence/datasets
- design/inverse-folding
Jul 28, 2026
PLMs are biased by uneven distribution of sequence data in datasets such as UniRef and UniProt
- evidence/datasets
- training/pretraining-and-scaling
Jul 28, 2026
Zero-shot performance of PLMs, but not inverse folding models, correlates with number of homologs available for training
- prediction/variant-effects
- evidence/datasets
Jul 20, 2026
AlphaFold3 inference benefits more from phylogenetically diverse MSAs than deep redundant MSAs
Jul 20, 2026
Heavy-chain templates and self-distillation improve nanobody structure prediction
Jul 17, 2026
Antibody-antigen complex prediction scales with training data and compute
Jul 17, 2026
Antibody-antigen modeling by diffusion-based structure prediction is data-limited
- prediction/complexes
- evidence/datasets
May 10, 2026
Model type less important than data quantity when predicting catalytic properties of protein sequences
- prediction/function
- evidence/datasets
Apr 21, 2026
AlphaFold3 uses cross-distillation from AlphaFold2 to avoid hallucinating secondary structure in low-pLDDT regions
- evidence/datasets
- prediction/structure
Apr 21, 2026
Alternate sequence clustering schemes outperform uniform sampling when training protein language models
- evidence/datasets
- training/pretraining-and-scaling
Apr 21, 2026
BERT models trained on sequence clusters outperform those trained on all data
- evidence/datasets
- training/pretraining-and-scaling
Apr 21, 2026
Composition of negative data affects rule discovery in binary classification tasks
Apr 21, 2026
Diffusion-based protein design methods undersample structural diversity in specific topologies
- design/backbones
- evidence/datasets
Apr 21, 2026
Distillation
- evidence/datasets
- training/pretraining-and-scaling
Apr 21, 2026
Diverse MSAs are better for training structure prediction neural networks than random MSAs
- evidence/datasets
- prediction/structure
Apr 21, 2026
Including sequences from multiplexed ancestral sequence reconstruction improves PLM training
Apr 21, 2026
Insufficient ddG data on Abs for training
- prediction/binding
- evidence/datasets
Apr 21, 2026
Inverse folding models trained on all proteins generate sequences that are equally thermostable as those generated by models trained only on thermophilic proteins
Apr 21, 2026
ML models trained exclusively on experimental structures are less effective on computational models
Apr 21, 2026
Masked PLMs are more sensitive to training imbalances than autoregressive PLMs
- evidence/datasets
- training/pretraining-and-scaling
Apr 21, 2026
Number of sequences with 90% homology is a greater predictor of viral variant effect prediction than alignment depth
Apr 21, 2026
Observable Antibody Space
- evidence/datasets
- antibodies/repertoires
Apr 21, 2026
PLMs are biased toward germline antibodies
- evidence/datasets
- antibodies/repertoires
Apr 21, 2026
Paired antibody LMs outperform equivalent unpaired LMs on sequence recovery for various regions
Apr 21, 2026
Protein backbone diffusion models undersample loop-rich and alpha-beta domains and functional motifs
- design/backbones
- evidence/datasets
Apr 21, 2026
Protein models designed using inverse folding can be used to supplement training DBs for PLMs and structure prediction models
- evidence/datasets
- training/pretraining-and-scaling
Apr 21, 2026
Self-distillation exposes more data to Evoformer
- evidence/datasets
- prediction/structure
Apr 21, 2026
Sequence clustering of training data affects variant effect prediction performance by PLMs
- evidence/datasets
- prediction/variant-effects
Apr 21, 2026
Sequence homology composition can affect performance of fine-tuned protein language models for variant effect prediction
Apr 21, 2026
Structure prediction neural networks trained on the same structural dataset are likely to predict the same conformation
- prediction/ensembles
- evidence/datasets
Apr 21, 2026
Training backbone diffusion models on synthetic data improves designability
- evidence/datasets
- design/backbones
Apr 21, 2026
Training on newly updated databases doesn't guarantee better performance
- evidence/datasets
- training/pretraining-and-scaling
Apr 21, 2026
Training protein structure prediction neural networks on both positive and negative protein-protein interactions improves PPI discrimination
- prediction/binding
- evidence/datasets
Apr 21, 2026
Unbalanced composition of sequence data prevents protein fitness from being identifiable from sequence data alone
Apr 21, 2026
A greater proportion of viral sequences are closely related to one another than prokaryotic and eukaryotic sequences, and these sequences get removed during clustering
- evidence/datasets
- evolution/selection
Apr 21, 2026
About 100k datapoints required to train accurate ddG predictor
- prediction/stability-expression
- evidence/datasets
Apr 21, 2026
Contrastive fine-tuning PLMs on inverse folding embeddings of experimental structures but not computational models improves downstream tasks
Apr 21, 2026
Focused protein sequence libraries are poor training sets
- design/directed-evolution
- evidence/datasets
Apr 21, 2026
Including crystal contacts in training data improves sidechain prediction
- prediction/structure
- evidence/datasets
Apr 21, 2026
Including nonfunctional sequences during training improves prediction of negatives but not positives
- evidence/datasets
- prediction/variant-effects
Apr 21, 2026
Incorporating ancestral sequences during PLM training improves performance in downstream tasks
Apr 21, 2026
Mega-scale dataset is inaccurate for DNA-binding domains
- evidence/measurements
- evidence/datasets
Apr 21, 2026
Models trained on synthetic data tend to converge to mean
- evidence/datasets
- evidence/generalization
Apr 21, 2026
Poisoning attacks on LMs require a constant number of samples regardless of scale
- evidence/generalization
- evidence/datasets
Apr 21, 2026
ProteinGym
- evidence/datasets
- prediction/variant-effects
Apr 21, 2026
Weighting training sequences by fitness improves identification of other fit sequences
- evidence/datasets
- prediction/variant-effects
Apr 20, 2026
Stability and thermostability
Apr 20, 2026
Top2018
- evidence/datasets
- prediction/structure

Created with Quartz v4.5.2 © 2026

GitHub
Discord Community