AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

3DCS: Datasets and Benchmark for Evaluating Conformational Sensitivity in Molecular Representations

Xi Wang, Yang Zhang, Yingjia Zhang, Yejia Cai, Shenji Wan

ICLR 2026 · Probabilistic and scientific modeling › Molecules and proteins · approach: Diffusion for molecular conformations

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
fine-tune with supervised contrastive loss on stereoisomersUniMol · 3DCS chiralityES-AUC: 0.622 -> 0.976Improved
fine-tune with supervised contrastive loss on stereoisomersMolAE · 3DCS chiralityES-AUC: 0.782 -> 0.971Improved
fine-tune with supervised contrastive loss on stereoisomersFMG · 3DCS chiralityES-AUC: 0.706 -> 0.999Improved
fine-tune with supervised contrastive loss on stereoisomersMACE · 3DCS chiralityES-AUC: 0.485 -> 0.887Improved
use cloze-style masked reconstruction instead of coordinate denoising pretrainingMolAE vs UniMol · 3DCS chirality (zero-shot)ES-AUC: 0.622 -> 0.782Improved
add chiral-aware field-based encoding (FMG) versus equivariant encodersFMG · 3DCS chirality (zero-shot)ES-AUC: 0.622 -> 0.706Improved
add energy and force supervision with spherical harmonic inputsMACE · 3DCS energySpearman: 0.026 -> 0.236Improved
use learned 3D encoder instead of E3FP fingerprintUniMol · 3DCS geometrySpearman: 0.406 -> 0.697Improved
use energy/force-trained encoder before RMD17 fine-tuningGemNet · Revised MD17average energy MAE (kcal/mol): 3.941 -> 0.371Improved
add spectra (MolSpectra) to 3D denoising pretrainingMolSpectra vs UniMol · 3DCS geometrySpearman: 0.697 -> 0.682No clear effect
use spectra-augmented pretraining on chiralityMolSpectra · 3DCS chirality (zero-shot)ES-AUC: 0.622 -> 0.545Hurt

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