Scholars Central

Conference abstractPublished 2019

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

Authors: Ruilin Li, Xinyin Han, Zhipeng He, Xiaoyu He, Shanyu Chen, Haidong Zhu, Yu Zhang, Dan Zhao, Weizhong Li, Beifang Niu

Chinese Academy of Sciences, Beijing, China University of Chinese Academy of Sciences, China J. Craig Venter Institute, USA

Ask about this article

Abstract

Select any words in the title or abstract, or press this and choose a sentence.

Gene prediction is an important approach to improve the annotation of metagenomic genes. A variety of gene prediction models based on different principles had been implemented, with emphasis on statistical models, Markov or improved Markov models, deep learning models, and so on. The current gene prediction algorithms, such as FragGeneScan, Prodigal, MetaGeneAnnotator, Orphelia, Glimmer3, GeneMarkS-2, were specially designed for short fragments or whole genomes; however, the former will result in the identified genes being incomplete and the latter is not suitable for unknown species. Meanwhile, according to our previous benchmark results of these algorithms, the prediction error rate was relatively high (27.10%~54.70%), especially for datasets with low coverage (staggered dataset). In this study, we proposed an algorithm based on feature selection of ORFs named as Consensus, which combined the ORFs generated from known models, extracted the ORFs’ feature matrix and the corresponding label matrix. Finally, the optimal solution was obtained by the least square’s solution of the feature and label matrixes. The overall indicator of gene prediction via Consensus was better than that of single software (F-score was 82.94% on staggered dataset). Even more remarkably, we compared the results of models using two longer assembled scaffolds datasets of the real mock metagenomic samples containing 20 bacterial strains from NCBI (National Center for Biotechnology Information) instead of simulated reads, which would truly reflect the predictive power of the models. We believe our findings will improve the study of novel genes and annotation pipelines in unknown metagenomic species.

Ask about this article

Answers come only from this article's title, abstract and keywords, each sentence pointing to where it comes from. Your question is sent to OpenAI to write the answer. Scholars Central does not store it.

Try:

More from this journal

The newest in Journal of Applied Bioinformatics & Computational Biology.

All its articles

Conferences in this field

Upcoming conferences in this field.

Follow

New articles from what you follow go into your email digest, once you agree to alert emails.

  • JournalJournal of Applied Bioinformatics & Computational Biology
  • SubjectBiochemistry & Biotechnology
  • SubjectMathematics & Statistics
  • BylineArticles whose byline prints “Ruilin Li”
  • BylineArticles whose byline prints “Xinyin Han”
  • BylineArticles whose byline prints “Zhipeng He”
  • BylineArticles whose byline prints “Xiaoyu He”
  • BylineArticles whose byline prints “Shanyu Chen”
  • BylineArticles whose byline prints “Haidong Zhu”
  • BylineArticles whose byline prints “Yu Zhang”
  • BylineArticles whose byline prints “Dan Zhao”

For authors and organisers

Publish, present and keep up with your field

Find the right journal for your paper and submit your manuscript.

Photo: Pexels