Abstract


N6-methyladenosine (m6A) is the most prevalent post-transcriptional modification in eukaryotes, and plays a pivotal role in various biological processes, such as splicing, RNA degradation and RNA-protein interaction. We propose here a prediction framework WHITSLE for transcriptome-wide m6A RNA methylation site prediction. When tested on six independent datasets, our approach, which integrated 35 additional genomic features besides the conventional sequence features, achieved a major improvement in the accuracy of m6A site prediction (average AUC: 0.960 and 0.895 under the full transcript or mature mRNA models, respectively) compared to the state-of-the-art computational approaches MethyRNA (AUC: 0.830 and 0.772) and SRAMP (AUC: 0.835 and 0.794). It also out-performed the existing epitranscriptome databases MeT-DB (AUC: 0.833 and 0.782) and RMBase (AUC: 0.822 and 0.775), which were built upon hundreds of epitranscriptome high-throughput sequencing samples. To unveil the putative biological processes impacted by changes in an individual m6A site, a network-based approach was implemented according to the ¡°guilt-by-association¡± principle by integrating the RNA methylation profile, gene expression profile and protein-protein interaction data. Finally, the WHISTLE web server was built to facilitate the query of the high-accuracy map of the human m6A epitranscriptome, and the server is available at: http://whistle-epitranscriptome.com

General results:

Model

Method

Testing on high resolution technique data£šAUC£©

m6A-CLIP

miCLIP

High resolution

m6A-seq

Average

Full

Transcript

WHISTLE

0.961

0.959

0.983

0.968

MethyRNA

0.821

0.838

0.918

0.859

SRAMP

0.840

0.828

0.824

0.831

RMBase

0.840

0.804

0.804

0.816

MetDB

0.851

0.816

0.805

0.824

Mature

mRNA

WHISTLE

0.919

0.872

0.917

0.903

MethyRNA

0.769

0.775

0.844

0.796

SRAMP

0.797

0.789

0.782

0.789

RMBase

0.787

0.763

0.789

0.780

MetDB

0.794

0.769

0.798

0.787



Citation


Please cite WHISTLE by including the following reference in your manuscript:

Kunqi Chen#, Zhen Wei#, Qing Zhang#, Xiangyu Wu#, Rong Rong, Zhiliang Lu, Jionglong Su, Joao Pedro de Magalhaes, Daniel Rigden and Jia Meng*. "WHISTLE: a functionally annotated high-accuracy map of the human N6-methyladenosine (m6A) epitranscriptome predicted from machine learning", Nucleic Acids Research, 2019.