Prodiction Of Protein Subschoroplast Locations Based On Different Features

The chloroplast is the main vital activity place of plant and algal cells,also are the main organelles of photosynthesis and energy interchange.Protein subschloroplast location prediction is a deeper problem than subcellular location,and it is significant to improve the accuracy of predictive subchloroplast locations for understanding their interactions with other molecules.So that becomes a new research of protein subcelluar localization.In our work,a new dataset of protein subchloroplast location with sequences'similarity less than 60%,called PS60,is constructed by searching for protein subschloroplast information of Swiss-Prot database,it includes three subcellular locations.We calculated three types of amino acid sequence information,including amino acid fragments compositional information,distant amino acid 2-dipeptide coupling information and amino acid index,two types of protein structure information,including predictive secondary structure and protein blocks information,gene ontology based on biological process and molecular function,and the evolutionary information as well as conservative information.The subchloroplast locations of proteins are predicted by using the algorithm of surpport vector machine.The overall prediction accuracy is 93.35%in the jackknife test.The better results are also obtained in the cross-validation and independent test,they are 93.72%and 90.65%,respectively.