生物医学データマイニングの国際ジャーナル
オープンアクセス

概要

Gene prediction algorithm based on feature selection of open reading frame for metagenomic sequences

Ruilin Li

Gene prediction is an important approach to to deal with improve the comment of metagenomic qualities. An assortment of quality expectation models dependent on various standards had been executed, with accentuation on measurable models, Markov or improved Markov models, profound learning models, etc. The current quality forecast calculations, for example, FragGeneScan,Prodigal, MetaGeneAnnotator, Orphelia, Glimmer3, GeneMarkS-2, were uniquely intended for short sections or entire genomes; notwithstanding, the previous will bring about the recognized qualities being fragmented and the last isn't reasonable for obscure species.

In the interim, as per our past benchmark aftereffects of these calculations, the expectation blunder rate was moderately high (27.10%~54.70%), particularly for datasets with low inclusion (staggered dataset). In this investigation, we proposed a calculation dependent on highlight choice of ORFs named as Consensus, which consolidated the ORFs created from known models, extricated the ORFs' component lattice and the comparing mark network. At long last, the ideal arrangement was acquired by the most un-square's answer of the element and mark grids. The general pointer of quality forecast through Consensus was superior to that of single programming (F-score was 82.94% on stunned dataset).