PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
February 3, 2021Bioinformatics1,338 citationsOpen Access

DNABERT: pre-trained Bidirectional Encoder Representations from Transformers model for DNA-language in genome

View Full Paper
YJYanrong JiZZZhihan ZhouHLHan Liu

Key Points

  • Develop DNABERT, a pre-trained bidirectional transformer framework designed to capture the syntax, semantics, and contextual representations of genomic DNA sequences.
  • Adapted the bidirectional encoder representations from transformers (BERT) architecture for genomic sequence processing.
  • Pre-trained the model on large-scale genome sequence data using nucleotide k-mer tokenization to capture bidirectional context.
  • Constructed a foundation model capable of learning universal representations of complex genomic sequence syntax.
  • Demonstrated effective feature transferability for downstream genomic prediction tasks and sequence interpretation.

Abstract

Supplementary data are available at Bioinformatics online.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Ji et al. (2021) studied this question.

synapsesocial.com/papers/69cd082cb6ea19ea46cbfbc6https://doi.org/10.1093/bioinformatics/btab083
Ask AI
Helpful
Bookmark
Share
View Full Paper