🧬 AbMelt: Antibody Thermostability Predictor

Predict antibody thermostability from molecular dynamics descriptors

This tool uses machine learning models trained on molecular dynamics simulations to predict three key thermostability properties:

Tagg: Aggregation temperature (higher = more aggregation prone)
Tm: Melting temperature (higher = more thermally stable)
Tmon: Melt onset temperature (higher = more stable)

Based on the AbMelt methodology from Rollins et al. (2024).

RMSF CDRs at 400K

Root Mean Square Fluctuation of CDR regions at 400K (typical range: 0.15-0.35)

Radius of Gyration Std CDRs at 400K

Standard deviation of radius of gyration for CDRs at 400K (typical range: 0.02-0.08)

All-Temperature Lambda Descriptor

Multi-temperature molecular dynamics descriptor (typical range: 0.8-1.6)

Thermostability Predictions

Examples

RMSF CDRs at 400K	Radius of Gyration Std CDRs at 400K	All-Temperature Lambda Descriptor

About AbMelt

AbMelt combines multi-temperature molecular dynamics simulations with machine learning to predict antibody thermostability. The method uses descriptors computed from MD trajectories at 300K, 350K, and 400K to capture the dynamic flexibility of antibody structures.

Key Features:

Predicts aggregation and melting temperatures
Uses physically meaningful MD descriptors
Outperforms sequence-based methods
Requires only antibody Fv structure

Reference: Rollins, Z.A., et al. "AbMelt: Learning antibody thermostability from molecular dynamics." Biophysical Journal (2024).

GitHub: MSDLLCpapers/AbMelt