AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Generative Models for Graph-Based Protein Design

NeurIPS 2019 · Probabilistic and scientific modeling › Molecules and proteins · approach: Protein representation learning · 209 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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
Structured Transformer vs SPIN2structure-conditioned sequence model · CATH Single chainperplexity: 12.61 -> 9.03Helped
condition on structure vs unconditional LSTMprotein language model · CATH Allperplexity: 17.12 -> 6.85Helped
increase LSTM hidden size 128 -> 512unconditional LSTM · CATH Allperplexity: 17.13 -> 17.13No clear effect
add orientation edge features to distances (ablation)Structured Transformer · CATH Allperplexity: 7.87 -> 6.85Helped
use backbone dihedrals instead of Cα angles as node features (ablation)Structured Transformer · CATH Allperplexity: 7.83 -> 6.85Helped
replace attention with PairMLP aggregation (ablation)Structured Transformer · CATH Allperplexity: 6.85 -> 6.55Helped
use rigid backbone features instead of flexible topological features (ablation)Structured Transformer · CATH Allperplexity: 11.51 -> 6.85Helped
Structured Transformer vs Rosetta fixbbfixed-backbone design · Single chain test setnative sequence recovery: 17.9 -> 27.6Helped
rigid vs flexible backbone features (ablation)Structured Transformer · miniprotein mutation effectsPearson RMixed

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