MechRNA: prediction of lncRNA mechanisms from RNA-RNA and RNA-protein interactions

Gawronski, Alexander; Uhl, Michaël; Zhang, Yajia; Lin, Yen-Yi; Niknafs, Yashar S.; Ramnarine, Varune Rohan; Malik, Rohit; Feng, Felix Y.; Chinnaiyan, Arul M.; Sahinalp, S. Cenk; Backofen, Rolf

doi:10.1101/285965

Cited by 7 publications

(7 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the other hand, stems (S) occur much often at the important lncRNA positions of interactive pairs than at the important lncRNA positions of noninteractive pairs. These suggest that open/unpaired lncRNA positions perhaps play more important roles in their interactions with proteins, and is consistent with several studies in the literature [4,47].…”

Section: Structural Components At Important Positions In Interactive and Non-interactive Pairssupporting

confidence: 92%

See 1 more Smart Citation

DeepLPI: a multimodal deep learning method for predicting the interactions between lncRNAs and protein isoforms

et al. 2021

View full text Add to dashboard Cite

Background Long non-coding RNAs (lncRNAs) regulate diverse biological processes via interactions with proteins. Since the experimental methods to identify these interactions are expensive and time-consuming, many computational methods have been proposed. Although these computational methods have achieved promising prediction performance, they neglect the fact that a gene may encode multiple protein isoforms and different isoforms of the same gene may interact differently with the same lncRNA. Results In this study, we propose a novel method, DeepLPI, for predicting the interactions between lncRNAs and protein isoforms. Our method uses sequence and structure data to extract intrinsic features and expression data to extract topological features. To combine these different data, we adopt a hybrid framework by integrating a multimodal deep learning neural network and a conditional random field. To overcome the lack of known interactions between lncRNAs and protein isoforms, we apply a multiple instance learning (MIL) approach. In our experiment concerning the human lncRNA-protein interactions in the NPInter v3.0 database, DeepLPI improved the prediction performance by 4.7% in term of AUC and 5.9% in term of AUPRC over the state-of-the-art methods. Our further correlation analyses between interactive lncRNAs and protein isoforms also illustrated that their co-expression information helped predict the interactions. Finally, we give some examples where DeepLPI was able to outperform the other methods in predicting mouse lncRNA-protein interactions and novel human lncRNA-protein interactions. Conclusion Our results demonstrated that the use of isoforms and MIL contributed significantly to the improvement of performance in predicting lncRNA and protein interactions. We believe that such an approach would find more applications in predicting other functional roles of RNAs and proteins.

show abstract

Section: Structural Components At Important Positions In Interactive and Non-interactive Pairssupporting

confidence: 92%

“…necessary to identify what other biological molecules it is able to interact with, especially proteins [3,4]. By interacting with proteins, lncRNAs could regulate the expression of genes, influence nuclear architecture and modulate the activity of proteins [5].…”

mentioning

confidence: 99%

DeepLPI: a multimodal deep learning method for predicting the interactions between lncRNAs and protein isoforms

et al. 2021

View full text Add to dashboard Cite

show abstract

“…We were also able to identify the discriminative features for identification of pre-miRNAs in two Drosophila species. In addition, our feature selection method can be applied to identify lncRNAs [100] and predict their target genes [101] and functions [102], [103].…”

Section: Discussionmentioning

confidence: 99%

An Improved Method for Identification of Pre-miRNA in Drosophila

Chen

Wang

2020

IEEE Access

View full text Add to dashboard Cite

Identification of microRNAs is important in studies of regulation of gene expression in many biologyical processes. In this study, we developed an improved method for identification of microRNAs in Drosophila. We used the iLearn, PyFeat, and Pse-in-One methods to extract the features and then used Max-Relevance-Max-Distance (MRMD2.0) and t-Distributed Stochastic Neighbour Embedding (t-SNE) to reduce dimension of the features and the random forest classifier in Weka to identify miRNAs. With this method, we found that the discriminative features for identification of pre-miRNAs were, in Drosophila melanogaster, the occurrences of G_GUG and C_AGU when the value of the feature vector was greater than 2, and in Drosophila pseudoobscura, the 4-tuple nucleotide composition and the occurrence of 4-length neighbouring nucleic acids when the value of the feature vector was less than 0.02. These vectors covered all compositional information or the frequency of bases. Classification results showed the classification accuracy was 95.7% and 93.6%, the precision rate was 95.8% and 93.6%, and the recall rate was 95.7% and 93.6% in Drosophila melanogaster and Drosophila pseudoobscura, respectively, which are higher than those reported in previous studies.

show abstract

“…This also allows GraphProt2 to calculate binding profiles, by using variable window sizes for computing position-wise binding scores. Binding profiles have proven to be of great practical value, and have been successfully applied in studies such as [7, 27]. In contrast, CNN methods are constrained to fixed-sized inputs.…”

Section: Introductionmentioning

confidence: 99%

GraphProt2: A graph neural network-based method for predicting binding sites of RNA-binding proteins

Uhl

Tran

Heyl

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

CLIP-seq is the state-of-the-art technique to experimentally determine transcriptome-wide binding sites of RNA-binding proteins (RBPs). However, it relies on gene expression which can be highly variable between conditions, and thus cannot provide a complete picture of the RBP binding landscape. This necessitates the use of computational methods to predict missing binding sites. Here we present GraphProt2, a computational RBP binding site prediction method based on graph convolutional neural networks (GCN). In contrast to current CNN methods, GraphProt2 supports variable length input as well as the possibility to accurately predict nucleotide-wise binding profiles. We demonstrate its superior performance compared to GraphProt and a CNN-based method on single as well as combined CLIP-seq datasets.Introduction 1 RNA-binding proteins (RBPs) regulate many vital steps in the RNA life cycle, such as 2 splicing, transport, stability, and translation [1]. Recent studies suggest a total number 3 of more than 2,000 human RBPs, including 100s of unconventional RBPs, i.e., RBPs 4 lacking known RNA-binding domains [2-4]. Numerous RBPs have been implicated in 5 diseases like cancer, neurodegeneration, and genetic disorders [5-7], urging the need to 6 speed up their functional characterization and shed light on their complex cellular 7 interplay. 8 An important step to understand RBP function is to identify the precise RBP 9 binding locations on regulated RNAs. In this regard, CLIP-seq (cross-linking and 10 immunoprecipitation followed by next generation sequencing) [8] together with its 11 popular modifications PAR-CLIP [9], iCLIP [10], and eCLIP [11] has become the 12state-of-the-art technique to experimentally determine transcriptome-wide binding sites 13 of RBPs. A CLIP-seq experiment for a specific RBP results in a library of reads bound 14 and protected by the RBP, making it possible to deduce its binding sites by mapping 15 the reads back to the respective reference genome or transcriptome. In practice, 16 computational analysis of CLIP-seq data has to be adapted for each CLIP-seq 17 protocol [12]. Within the analysis, arguably the most critical part is the process of peak 18 calling, i.e., to infer RBP binding sites from the mapped read profiles. Among the many 19 existing peak callers, some popular tools are Piranha [13], CLIPper [14], 20 PEAKachu [15], and PureCLIP [16].While peak calling is essential to separate authentic binding sites from unspecific 22 interactions and thus reduce the false positive rate, it cannot solve the problem of 23 expression dependency. In order to detect RBP binding sites by CLIP-seq, the target 24 RNA has to be expressed at a certain level in the experiment. Since gene expression 25 naturally varies between conditions, CLIP-seq data cannot be used directly to make 26 condition-independent binding assumptions on a transcriptome-wide scale. Doing so 27 would only increase the false negative rate, i.e., marking all non-peak regions as 28 non-binding, while in fact one cannot tell from the dat...

show abstract

MechRNA: prediction of lncRNA mechanisms from RNA-RNA and RNA-protein interactions

Cited by 7 publications

References 59 publications

DeepLPI: a multimodal deep learning method for predicting the interactions between lncRNAs and protein isoforms

DeepLPI: a multimodal deep learning method for predicting the interactions between lncRNAs and protein isoforms

An Improved Method for Identification of Pre-miRNA in Drosophila

GraphProt2: A graph neural network-based method for predicting binding sites of RNA-binding proteins

Contact Info

Product

Resources

About