Protein structure prediction is very heated research area in computational bioinformatics field. A Faster and more accurate protein secondary structure prediction tool can provide important support for ab initio and homology modelling protein 3D structure prediction. The protein secondary structure is acting as a bridge that link the sequence and 3D structure.
Citation: Fang, Chao, Yi Shang, and Dong Xu. "MUFold-SS: Protein Secondary Structure Prediction Using Deep Inception-Inside-Inception Networks." arXiv preprint arXiv:1709.06165 (2017).
Citation: Fang, Chao, Yi Shang, and Dong Xu. "MUFold-SS: Protein Secondary Structure Prediction Using Deep Inception-Inside-Inception Networks." PROTEINS: Structure, Function, and Bioinformatics (2018).
Please note that MUFold-SS is free for academic use purpose only, for commercial usage, please contact authors(See contact page)
1. You need to have a Psiblast database to search for profile. The MUFold-SS model was trained using profiles which were generated by Psiblast using the UniRef50 database. This database is still large, so when we benchmark the tool using test sets. We use a smaller database, UniRef50_Shrinked. We filtered the database so that it contains proteins whose length falls into range[70,3000]. For more database, please see Database Menu
2. You need to download the psiblast toolkit from NCBI Psiblast offical website
1. download the MUFold-SS stand alone package
2. download and install psiblast
3. install Tensorflow and Keras
4. follow the configuration file in the stand alone package and run