EnhancerP-2L: A Gene regulatory site identification tool for DNA enhancer region using CREs motifs

Abstract

Enhancers are DNA fragments that do not encode RNA molecules and proteins, but they act critically in the production of RNAs and proteins by controlling gene expression. Prediction of enhancers and their strength plays significant role in regulating gene expression. Prediction of enhancer regions, in sequences of DNA, is considered a difficult task due to the fact that they are not close to the target gene, have less common motifs and are mostly tissue/cell specific. In recent past, several bioinformatics tools were developed to discriminate enhancers from other regulatory elements and to identify their strengths as well.However the need for improvement in the quality of its prediction method requires enhancements in its application value practically. In this study, we proposed a new method that builds on nucleotide composition and statistical moment based features to distinguish between enhancers and non-enhancers and additionally determine their strength. Our proposed method achieved accuracy better than current state-of-the-art methods using 5-fold and 10-fold cross-validation. The outcomes from our proposed method suggest that the use of statistical moments based features could bear more efficient and effective results. For the accessibility of the scientific community, we have developed a user-friendly web server for EnhancerP-2L which will increase the impact of bioinformatics on medicinal chemistry and drive medical science into an unprecedented resolution. Web server is freely accessible at http://www.biopred.org/enpred.

Enhancer-LSTMAtt: A Bi-LSTM and Attention-Based Deep Learning Method for Enhancer Recognition

“…In this case, the overall indices could be used, such as ACC and MCC. In the 5-fold cross-validation, Enhancer-LSTMAtt was superior to Enhancer-BERT [ 55 ], DeployEnhancer [ 48 ] and iEnhancer-RF [ 57 ] in terms of ACC and MCC in the first stage and exceeded iEnhancer-PsedeKNC [ 41 ], DeployEnhancer [ 48 ], EnhancerP-2L [ 51 ], and iEnhancer-RF [ 57 ] in terms of MCC in the second stage. In the 10-fold cross-validation, Enhancer-LSTMAtt reached competitive performance with ES-ARCNN [ 49 ], iEnhancer-XG [ 53 ], and iEnhancer-MFGBDT [ 63 ] in the second stage.…”

“…For fair comparison with the state-of-the-art methods, we used the same benchmark dataset as those in iEnhancer-2L [ 40 ], iEnhancer-PsedeKNC [ 41 ], EnhancerPred [ 42 ], EnhancerPred2.0 [ 43 ], Enhancer-Tri-N [ 44 ], iEnhaner-2L-Hybrid [ 45 ], iEnhancer-EL [ 46 ], iEnhancer-5Step [ 47 ], DeployEnhancer [ 48 ], ES-ARCNN [ 49 ], iEnhancer-ECNN [ 50 ], EnhancerP-2L [ 51 ], iEnhancer-CNN [ 52 ], iEnhancer-XG [ 53 ], Enhancer-DRRNN [ 54 ], Enhancer-BERT [ 55 ], iEnhancer-KL [ 56 ], iEnhancer-RF [ 57 ], spEnhancer [ 58 ], iEnhancer-EBLSTM [ 59 ], iEnhancer-GAN [ 60 ], piEnPred [ 61 ], iEnhancer-RD [ 62 ], and iEnhancer-MFGBDT [ 63 ]. The dataset was initially collected by Liu et al [ 40 ] from chromatin state information of nine cell lines (H1ES, K562,GM12878, HepG2, HUVEC, HSMM, NHLF, NHEK and HME) which was annotated by ChromHMM [ 69 , 70 ].…”

“…Furthermore, Machine learning-based methods are capable of discovering non-linear hidden motifs of enhancers. To date, there are at least twenty machine learning-based methods for enhancer prediction [ 16 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 , 61 , 62 , 63 ], such as iEnhancer-2L [ 40 ], iEnhancer-PsedeKNC [ 41 ], EnhancerPred [ 42 ], and EnhancerPred2.0 [ 43 ]. The general workflow of these methods is firstly to compute representation of sequences such as pseudo k-tuple nucleotide composition, nucleotide binary profiles, as well as accumulated nucleotide frequency, then to learn a classifier by using a machine learning algorithm such as support vector machine and random forest, and finally to predict unknown sequences.…”

Fast activation maximization for molecular sequence design

“…Specifically, when designing polyadenylation signals based on APARENT, we validated the designs using DeeReCT-APA [31], an LSTM trained on 3'-sequencing data of mouse cells, and DeepPASTA [30], a CNN trained on human 3'-sequencing data. When designing enhancer sequences, we validated the designs using iEnhancer-ECNN [62], an ensemble of CNNs trained on genomic enhancer sequences, and EnhancerP-2L [63], a Random Forest-classifier based on statistical features extracted from enhancer regions in the genome. Finally, to validate Optimus 5' designs, we had access to a newer version of the model that had been trained on additional MPRA data, making it more robust particularly on outlier sequences such as long homopolymer stretches [25].…”

“…EnhancerP-2L [63] Detects genomic enhancer regions and predicts whether it is a weak or strong enhancer. For a sample of generated sequences per design method, we calculated the mean detect/not detect prediction rate, the mean weak/strong prediction rate and the mean p-score.…”