Prediction Of Protein Submitochondria Locations Using Protein And MRNA Sequence Information

Mitochondria is a kind of important semi-autonomous organelles,it mainly provides energy for the cell and bears many other physiological functions.But the some roles of mitochondria are performed only in a particular submitochondria location.In this study,a dataset M254 with mRNA sequences is constructed,an approach for predicting protein submitochondria locations is proposed by combining evolutionary information,gene ontology of homologous sequences,amino acid indices,stickiness,amino acid composition,the average chemical shift and the frequency of mRNA.The overall prediction accuracy is 97.64%for the submitochondria location by using the algorithm of the support vector machine and increment of diversity in jackknife test.For the independent test,the good prediction results were also obtained.In order to find the more effective information parameter for predicting the submitochondria locations of proteins,another dataset M880 included more proteins with mRNA sequence is constructed.By combining gene ontology of homologous sequences,amino acid indices,amino acid composition,evolutionary information,the frequency of mRNA triplet code as well as the triplets of structure-sequence mode based on the kinds of base and mRNA secondary structure,the submitochondria locations of proteins are predicted by using the algorithm of the support vector machine.The overall prediction accuracy is 96.02%and 95.45%,respectively,in jackknife test and independent test.These results show that our parameter selection and algorithm are more effective.