ABSTRACT
MViewEMA, a single-model EMA method that leverages a multi-view representation learning framework to integrate residue–residue interaction features from micro-environment, meso-environment, and macro-environment levels for global accuracy assessment of protein complex models. Benchmark results demonstrate that MViewEMA outperforms state-of-the-art EMA methods in global accuracy assessment, achieving over 10-fold improvement in computational efficiency compared to our previous method, DeepUMQA3. Integrating MViewEMA into modern prediction frameworks, such as AlphaFold-Multimer, AlphaFold3, and DiffDock-PP, can enhance the accuracy of complex structure prediction. This method facilitates efficient selection of high-quality protein complex models from large-scale structural datasets and demonstrated top performance in model selection tracks during the CASP16 blind test.
News
References