A novel comparative sequence analysis method for ncRNA secondary structure prediction without multiple sequence alignment

Consensus RNA secondary structure and helices are always required from several homologous sequences as the development of non-coding RNA. The problem can be solved by comparative sequence analysis methods, in which nearly all the algorithms are based on multiple sequence alignment. However, there are no effective models combining similarity and secondary structure. So far multiple sequence alignment is an open problem, especially applied in predicting RNA secondary structure. In this paper, we propose a novel comparative sequence method without multiple sequence alignment. It compares the position of the helices in dot plots matrix instead of sequences. The concept 'centorid of helix' is presented and a novel algorithm is designed for finding consensus helices. Experiments on tRNA prove that our algorithm outperforms the current main software, including Pfold, MARNA, CARNAC and RNAalifold.