Identification of non-coding RNAs in Mycobacterium Tuberculosis genome using combined computational approach

Abstract

Nowadays it has already been known that non-coding RNAs (ncRNAs), which are not translated to proteins, play important roles in cellular processes including regulatory functions. In order to identify putative ncRNAs of Mycobacterium tuberculosis, genome-wide screening by using computational approach is applied. Although the efficiency of currently available programs is limited, combined approach was the method of choice. New workflow development was required. The core program, RNAz, of the workflow was integrated with TBA. By testing the workflow with Escherichia coli genome, it was however, observed that TBA generated a large number of false positives by generating missing alignment. This problem is challenging. In order to solve this, new genome wide alignment protocol was developed by combining BLAST search and MAFFT multiple sequence alignment. Evaluating this with E. coli ncRNA prediction, it can improve sensitivity of RNAz results from 0.54 to 0.84, precision from 0.37 to 0.56 and reduce time to calculate from over 6 hours to 70 minutes. Therefore, this protocol was used, instead of TBA, in M. tuberculosis ncRNA gene identification, resulting 61 predicted loci. Based on M. tuberculosis H37Rv ncRNA annotation, 33 predicted RNA loci were located in ncRNA gene region. Other loci were mapped with promoter and terminator prediction. There were 22 loci which had transcription signal and only a locus had double transcription signal. By sequence similarity search, there were 3 loci which matched with two known RNA sequences, ykoK and SRP, in database. The ykoK element is a reglatory element of divalent cation-related genes and SRP involves in protein translocation in cell. Resuling candidate putative loci were considered as putative ncRNAs for further experimental verification.