Back

Peak2Vec Enables Inferrence Of Transcriptional Regulation From Atac-Seq

Liang, L.; Lu, X.; Lu, S.

2021-10-01 bioinformatics
10.1101/2021.09.29.462455 bioRxiv
Show abstract

Transcription factor (TF) binding sites in ATAC-seq are typically determined by footprint analysis. However, the performance of footprint analysis remains unsatisfying and most TFs do not exhibit footprint patterns. In this study, we modified the convolutional neural network to project sequences into an embedding space. Sequences with similar nucleotide patterns will stay close together in the embedding. The dimensionality of this embedding space represents binding specificities of various TFs. In the simulation experiment, peak2vec accurately distinguished the three TFs in the embedding space while conventional deep learning cannot. When applied to the ATAC-seq profiles of hepatitis carcinoma, peak2vec recovered multiple motifs curated in database, while significant portion of sequences corresponding to the TF are located at the promoter region of its regulated genes.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.