AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

ProtDBench: A Unified Benchmark of Protein Binder Design and Evaluation

Cong Liu, Milong Ren, Jiaqi Guan, Chengyue Gong, Jinyuan Sun, Xinshi Chen, Wenzhi Xiao

ICML 2026 · Probabilistic and scientific modeling › Molecules and proteins · approach: Generative protein structure design · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use diffusion PXDesign instead of hallucination BindCraftbinder design generation · PDL1successful backbones per 24h: 231 -> 1247Improved
use diffusion PXDesign instead of hallucination BindCraftbinder design generation · BHRF1successful backbones per 24h: 122 -> 837Improved
use PXDesign instead of RFDiffusion-3binder design generation · PDL1successful backbones per 24h: 542 -> 1247Improved
use BindCraft instead of PXDesignbinder design generation · Cao PDGFRper-sequence success rate: 29.80 -> 54.19Improved
use BindCraft instead of PXDesignbinder design generation · Cao VirB8per-sequence success rate: 45.82 -> 16.04Hurt
use hallucination BindCraft instead of PXDesignbinder design generation · Cao PDGFRcluster pass rate: 0.52 -> 48.68Improved
swap Protenix for lightweight Protenix-Mini as verifier (ablation)evaluation verifier · Cao combined filterssuccess rateNo clear effect
replace single verifier with union of Top-1% ipTM across verifiers (ablation)evaluation verifier · Cao recallunion recallImproved
require agreement of at least k verifiers (ablation)evaluation verifier · Cao recallrecallHurt
use stricter Protenix-Mini verifier instead of AF2-IG-Easy (ablation)generative benchmark · 10 targetssuccess rateHurt
use two-metric filter (binder pTM, binder ipTM) instead of triplet (ablation)Protenix combined filter · Cao targetssuccess rateNo clear effect
restrict to all-alpha instead of mainly-beta binders (ablation)generative methods · 10 targetsAF2/Protenix metricsImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy