The Protein Model Quality Assessment

Welcome to GraphCPLMQA

GraphCPLMQA: Assessing protein model quality based on deep graph coupled networks using protein language model.

In GraphCPLMQA, sequence and structural embeddings are extracted by protein language models, and supplemented with structural features to predict model quality through the coupled neural networks. The network consists of a graph module that encodes the sequence-structure relationship and a transform-based convolutional module that decodes structure- quality the relationship.

GraphCPLMQA contains a single-sequence version of the GraphCPLMQA-Single method and an MSA version of the GraphCPLMQA-MSA method to assess protein model quality.


Evaluate structure using MSA language model.


Evaluate structures using a single sequence language model, fast.

Model Ordering

Sort multiple candidate models.

GraphCPLMQA papers