RBPSpot: Learning on appropriate contextual information for RBP binding sites discovery

Sharma, Nitesh Kumar; Gupta, Sharad; Kumar, Ashwani; Kumar, Prakash P.; Pradhan, Upendra Kumar; Shankar, Ravi

doi:10.1016/j.isci.2021.103381

Cited by 9 publications

(14 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The majority of methods use one-hot encoding to project the RNA sequence into a machine-readable format. However, several methods, including RBPSpot [54], iDeepV [55] and deepRAM [49], use a word2vec [56] model to first learn an embedding of nucleotide 3-mers in an unsupervised manner. During training, k-mers of the input RNA sequence are projected into the word2vec model's embedding, which then serves as input to subsequent layers.…”

Section: Deep Learning Architectures For Protein-rna Interaction Pred...mentioning

confidence: 99%

A Systematic Benchmark of Machine Learning Methods for Protein-RNA Interaction Prediction

Horlacher

Cantini

Hesse

et al. 2023

Preprint

View full text Add to dashboard Cite

RNA-binding proteins (RBPs) are central actors of RNA post-transcriptional regulation. Experiments to profile binding sites of RBPs in vivo are limited to transcripts expressed in the experimental cell type, creating the need for computational methods to infer missing binding information. While numerous machine-learning based methods have been developed for this task, their use of heterogeneous training and evaluation datasets across different sets of RBPs and CLIP-seq protocols makes a direct comparison of their performance difficult. Here, we compile a set of 37 machine learning (primarily deep learning) methods for in vivo RBP-RNA interaction prediction and systematically benchmark a subset of 11 representative methods across hundreds of CLIP-seq datasets and RBPs. Using homogenized sample pre-processing and two negative-class sample generation strategies, we evaluate methods in terms of predictive performance and assess the impact of neural network architectures and input modalities on model performance. We believe that this study will not only enable researchers to choose the optimal prediction method for their tasks at hand, but also aid method developers in developing novel, high-performing methods by introducing a standardized framework for their evaluation.

show abstract

Section: Deep Learning Architectures For Protein-rna Interaction Pred...mentioning

confidence: 99%

A Systematic Benchmark of Machine Learning Methods for Protein-RNA Interaction Prediction

Horlacher

Cantini

Hesse

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

Section: Introductionmentioning

confidence: 99%

“…The existing software are overtly dependent upon the old school of motif discovery and user defined motifs, while recent reports suggest that TF binding is more about context and surroundings ( Sharma et al 2021; Heikham and Shankar 2010; Zhou et al 2013 ). This step itself is heavily dependent upon binding experiments like PBM and ChIP-seq, on whose results, binding motif is define for any given TF.…”

Section: Introductionmentioning

confidence: 99%

“…With this all foundations and understanding of the challenges, we here present an approach and software, PTFSpot, to detect transcription factors binding regions which is based on the following principles: Instead of overtly relying on motifs, identify the most significant motifs specific to any transcription factor and use it as the seed to identify most significant flanking regions for additional information. The same theory we had successfully applied in discovering RNA binding proteins binding sites ( Sharma et al 2021 ). Build upon more realistic datasets instead of general practice of using random datasets and randomized sequences with similar sequence composition. Use the significant motifs found in ChIP-seq and DAP-seq data (positive datasets) to locate them in unbound regions for the given experimental condition.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

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

Gupta,

Kesarwani,

Bhati

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Unlike animals, variability in transcription factors (TF) and their binding sites (TFBS) across the plants is a major problem which most of the existing TFBS finding software fail to tackle, rendering them hardly of any use. This limitation has resulted into underdevelopment of plant regulatory research and rampant misuse ofArabidopsismodels, generating misleading results. Here we report a ground-breaking transformers based deep-learning approach, PTFSpot, which learns from TF structures and their binding sites to bring a universal TF-DNA interaction model. During a series of extensive bench-marking studies, it not only outperformed the existing software by >20% lead, but also delivered consistently >90% accuracy even for those species and TF families which were never encountered during model building process. PTFSpot makes it possible now to accurately annotate TFBS across novel plant genomes even in the total lack of any TF information.

show abstract

“…Moreover, these approaches completely ignore the importance of context and structural information, which have now been recognized as essential [45].…”

Section: Introductionmentioning

confidence: 99%

Comprehensive evaluation of plant transcription factors binding sites discovery tools

Ritu,

Gupta,

Shankar

2023

Preprint

Self Cite

View full text Add to dashboard Cite

Transcription factors (TFs) are regulatory proteins that bind to a specific DNA sequence, known as the transcription factor binding site (TFBS), to regulate the rate of transcription process. For a better understanding of these regulatory proteins, it is crucial to identify these TFBSs across the genomic sequences. The identification of TFBS has been made possible by a number of experimental and computational techniques established during the past few decades. These existing approaches are skewed heavily around human TFBS discovery. The area of TFBS identification in plants is lagging behind due to a lack of experimental data for TFs and very scarce attempts to build plant-specific TFBS identification models. Therefore, to facilitate a better understanding of tools to identify TFBS in plants, here we provide a comprehensive survey of a large number of TFBS identification tools. We have also provided a repository along with corrected scripts, which also has datasets used in the present study. We expect this study to serve as a guide to better the understanding of plant TFBS tools and the care to be taken while applying them, especially during cross-species applications. The corrected scripts of these software are made available athttps://github.com/SCBB-LAB/Comparative-analysis-of-plant-TFBS-software.

show abstract

RBPSpot: Learning on appropriate contextual information for RBP binding sites discovery

Cited by 9 publications

References 43 publications

A Systematic Benchmark of Machine Learning Methods for Protein-RNA Interaction Prediction

A Systematic Benchmark of Machine Learning Methods for Protein-RNA Interaction Prediction

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

Comprehensive evaluation of plant transcription factors binding sites discovery tools

Contact Info

Product

Resources

About