Model-DMS_DLG4_RAT-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 trained on a sigle site deep mutation scanning dataset and can be used to predict fitness score of mutant amino acid sequence of protein DLG4_RAT (Disks large homolog 4).

Protein Function

Postsynaptic scaffolding protein that plays a critical role in synaptogenesis and synaptic plasticity by providing a platform for the postsynaptic clustering of crucial synaptic proteins. Interacts with the cytoplasmic tail of NMDA receptor subunits and shaker-type potassium channels. Required for synaptic plasticity associated with NMDA receptor signaling. Overexpression or depletion of DLG4 changes the ratio of excitatory to inhibitory synapses in hippocampal neurons. May reduce the amplitude of ASIC3 acid-evoked currents by retaining the channel intracellularly. May regulate the intracellular trafficking of ADR1B.

Task type

protein level regression

Dataset description

The dataset is from Deep generative models of genetic variation capture the effects of mutations. And can also be found on SaprotHub dataset.

Label means fitness score of each mutant amino acid sequence. Ranging from negative infinity to positive infinity. If the effect larger than 0 represents high fitness, smaller than 0 represents low fitness.

Model input type

Amino acid sequence

Performance

0.70 Spearman's ρ

LoRA config

lora_dropout: 0.0

lora_alpha: 16

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

modules_to_save: ["classifier"]

Training config

class: AdamW

betas: (0.9, 0.98)

weight_decay: 0.01

learning rate: 1e-4

epoch: 50

batch size: 128

precision: 16-mixed