1
Change
Publication volume is up strongly over the prior window
The topic has 624 eligible papers in the recent 30-day window versus 323 in the prior 30-day window, an increase of 301 papers, or about 1.9×.
Window countsWeekly Trend BriefEvidence window ending 2026-05-18
Computational Drug Discovery
Recent output nearly doubled versus the prior 30-day window, led by papers on QSAR, toxicity prediction, drug–target interaction prediction, graph models, and virtual screening assessment.
Scollr found 624 eligible papers in Computational Drug Discovery Methods during the most recent 30 days, compared with 323 in the prior 30 days. The representative set is method-heavy: QSAR feature ensembling, conformal/Bayesian toxicity prediction, transformer and graph-attention drug–target prediction, binding-affinity modeling, graph-based drug–drug interaction reviews, and a critical assessment of binding-prediction tools. Because the representative papers are all 2026 items with zero recorded citations in the packet, and two are from preprint servers, the signal should be read as fast-moving research activity rather than settled evidence.
624Recent eligible papers
323Prior 30-day eligible papers
623Recent commentary-eligible papers
8Representative papers
1Recent abstractless papers
Current windowRecent eligible papers
ComparisonPrior eligible papers
Brief typeWeekly research trend
Evidence-backed changes
What's moving
2
Change
The available trend structure is broad rather than subdivided
The packet lists the top descendant as the parent topic itself, Computational Drug Discovery Methods, with 624 eligible papers and 623 commentary-eligible papers. The evidence threshold passes the recent-paper and representative-paper thresholds but does not pass the cluster threshold, so subtopic conclusions should be treated cautiously.
Top topic and quality-gate context3
Change
Current representative work centers on predictive modeling for compounds, targets, interactions, and screening
The representative papers cover QSAR prediction, toxicity prediction, chemical twin detection for virtual screening, compound–target and drug–target interaction prediction, drug–target binding affinity prediction, drug–drug interaction prediction, and assessment of binding-prediction tools.
Representative paper titles4
Change
Treat the newest findings as early signals
Every representative paper in the packet is from 2026 and has zero recorded citations and zero recorded FWCI. Two representative items are from ChemRxiv and bioRxiv, so their findings should not be presented as settled evidence.
Representative paper metadata
Topic shape
Theme clusters
Computational Drug Discovery Methods
The only top cluster in the packet is the parent topic itself. Recent representative papers span QSAR, toxicity prediction, drug–target and compound–target modeling, binding affinity, drug–drug interactions, and virtual screening, indicating a broad methods trend rather than a narrower subcluster resolved by the packet.
Parent-topic cluster and representative papersPredictive interaction and binding models
Several representative papers focus on predicting molecular interactions or binding outcomes: a review of compound–target interaction prediction, transformer/graph-attention drug–target interaction prediction, transformer-plus-embedding binding-affinity prediction, graph-based drug–drug interaction prediction, and a critical assessment of small-molecule binding prediction for virtual screening.
Representative interaction-model papers
Evidence anchors
Representative papers
Computational Drug Discovery Methods
Improving QSAR model predictions using ensembled heterogenous features
Representative of QSAR-modeling work in the recent window; note that the source is ChemRxiv and the packet records zero citations.
Computational Drug Discovery Methods
Mitigating silent failures in toxicity prediction: a conformalized heteroscedastic Bayesian framework
Shows attention to uncertainty and failure modes in toxicity prediction within the recent representative set.
Computational Drug Discovery Methods
TwinSAR: An Adaptive Kernel-based Algorithm with logit-transformed Z-score Filtering for Chemical Twin Detection in Large-scale Virtual Screening
Represents virtual-screening methodology; treat as preliminary because the source is bioRxiv and the packet records zero citations.
Computational Drug Discovery Methods
Machine Learning Approaches for Compound–Target Interaction Prediction: A Review
A review item in the representative set, useful for orienting readers to machine-learning approaches for compound–target interaction prediction.
Computational Drug Discovery Methods
TransGAT-DTI: Transformer and Graph Attention Network for Drug–Target Interaction Prediction
Representative of transformer and graph-attention architectures applied to drug–target interaction prediction.
Computational Drug Discovery Methods
TSEDTA: A Transformer-based neural network with SMILES Transformer and ESM2 embeddings for drug-target binding affinity prediction
Highlights transformer-based binding-affinity prediction using molecular and protein embeddings in the recent set.
Computational Drug Discovery Methods
Graph-based deep learning for drug-drug interaction prediction: a systematic review
A systematic review on graph-based deep learning for drug–drug interaction prediction, indicating review-level consolidation around graph methods.
Computational Drug Discovery Methods
The Last Mile Problem: A Critical Assessment of Physics-Based and AI Tools for Small Molecule Binding Prediction in Virtual Screening
Provides a critical-assessment angle on physics-based and AI binding-prediction tools for virtual screening.