Using random forest based on codon usage for predicting Human Leukocyte Antigen gene

摘要

Predicting of Human Leukocyte Antigen (HLA) gene can provide procedure into the human immune system. The classification of HLA genes has been developed by using various computational methods random forest based on codon usage. And ten-fold cross-validation to evaluate the models. Here, we propose methods of amino acid composition (AAC), dipeptide compositions (DPC) and p-collocated to investigate for major class/sub class HLA genes and to achieve high accuracy 96.24%, 98.25% and 99.25%, respectively, compared with the existing method. Finally, we shown nucleotide triplets code for a specific amino acid affect to predicting HLA gene.