SaProtHub/Model-Mega-sacle-Protein-Stability-Prediction-35M

Base model: westlake-repl/SaProt_35M_AF2

Model Card for Model ID

<!-- Provide a quick summary of what the model is/does. --> This model is used to predict protein stability (ΔΔG) for mutant amino acid sequence.

Task type

protein level regression

Dataset description

The dataset is from Mega-scale experimental analysis of protein folding stability in biology and design. We collect all protein sequences that have ΔΔG value.

Label is the ΔΔG (kcal/mol) value, the positive value means stable and the negetive value represents unstable, ranging from minus infinity to positive infinity.

Model input type

Amino acid sequence

Performance

test_loss: 0.18

test_spearman: 0.92

LoRA config

lora_dropout: 0.0

lora_alpha: 16

target_modules: ["query", "key", "value", "intermediate.dense", "output.dense"]

modulestosave: ["classifier"]

Training config

class: AdamW

betas: (0.9, 0.98)

weight_decay: 0.01

learning rate: 1e-4

epoch: 20

batch size: 64

precision: 16-mixed