PlantBind: an attention-based multi-label neural network for predicting plant transcription factor binding sites

Abstract

Identification of transcription factor binding sites (TFBSs) is essential to understanding of gene regulation. Designing computational models for accurate prediction of TFBSs is crucial because it is not feasible to experimentally assay all transcription factors (TFs) in all sequenced eukaryotic genomes. Although many methods have been proposed for the identification of TFBSs in humans, methods designed for plants are comparatively underdeveloped. Here, we present PlantBind, a method for integrated prediction and interpretation of TFBSs based on DNA sequences and DNA shape profiles. Built on an attention-based multi-label deep learning framework, PlantBind not only simultaneously predicts the potential binding sites of 315 TFs, but also identifies the motifs bound by transcription factors. During the training process, this model revealed a strong similarity among TF family members with respect to target binding sequences. Trans-species prediction performance using four Zea mays TFs demonstrated the suitability of this model for transfer learning. Overall, this study provides an effective solution for identifying plant TFBSs, which will promote greater understanding of transcriptional regulatory mechanisms in plants.

The bHLH transcription factor OsPRI1 activates the Setaria viridis PEPC1 promoter in rice

“…Transcription factors from the bHLH family are known to bind both E‐boxes and N‐boxes (Li et al ., 2006), but little is known regarding the nucleotide preference of each protein, with new softwares emerging to predict TF targets (e.g. Yan et al ., 2022; Cheng et al ., 2023). The S. viridis PEPC1 promoter sequence contains seven different E‐Boxes and two different N‐Boxes variants, making a total of 17 putative bHLH binding sites (Fig.…”

Comprehensive analysis of computational approaches in plant transcription factors binding regions discovery

“…Additionally, we used a large volume of DAP-seq data for Arabidopsis thaliana to fairly evaluate different software for the identification of TFBS in plants. This DAP-seq data was previously utilized by tools like TSPTFBS [51], PlantBind [53], and TSPTFBS 2.0 [63]. This dataset covered the peak data for 35 plant transcription factor families, as shown in Figure 3 with (a) a line plot and (b) a pie chart.…”

PTFSpot: Deep co-learning on transcription factors and their binding regions attains impeccable universality in plants

“…Many TFs share a similar binding motif but yet differ in their binding due to local surroundings and contexts ( Figure 1a & b ). Factors like DNA shape also play an important role in determining the binding sites for the TFs as shown very recently by some, including couple of software implementing shape information and observing performance improvement ( Mejía-Guerra and Buckler 2019; Ji et al 2021; Yan et al 2022; Sielemann et al 2021 ). Further to this, the choice of negative datasets with most of the software has been very relaxed as they randomly pick sequences and give too much weight to the consensus motif which can actually occur even in the non binding regions, creating weak datasets on which learnings have been done so far.…”