Table 5 Performance Comparison with other Methods in literature (with different dataset used in the proposed work).
From: Sequence based prediction of enhancer regions from DNA random walk
Authors | Datasets used | Features used | Method used | AUC/Accuracy (Acc) |
|---|---|---|---|---|
Bu, H., Gan, Y., Wang, Y., Zhou, S., & Guan, J.25 EnhancerDBN25 | Histone modification | DNA sequence compositional features, DNA methylation (GC content and DNA methylation) | Deep Belief Network | Acc 92.0% |
Yang, B., Liu, F., Ren, C., Ouyang, Z., Xie, Z., Bo, X., & Shu, W.8 BiRen8 | Human and mouse noncoding fragments in the VISTA Enhancer Browser | DNA sequence alone | Deep-learning-based hybrid architecture that integrates a Convolutional Neural Network (CNN) and a GRU-BRNN | AUC 0.956 |
Histone modifications (ChIPSeq), TFs and cofactors (ChIP-Seq), chromatin accessibility (DNase-Seq), transcription (RNA-Seq), DNA methylation (RRBS), CpG islands, evolutionary conservation, sequence signatures, and occupancy of TFBS. | 1,114-dimensional heterogeneous features in H1 cells, 22 training cell types/tissues | Deep learning | Acc 97.65% | |
Kim, S. G., Harwani, M., Grama, A., & Chaterji, S.11. EP-DNN11 | Chromatin features | p300 binding sites, as enhancers, and TSS and random non-DHS sites, as non-enhancers. We perform same-cell and cross-cell predictions to quantify the validation rate and compare against two state-of-the-art methods, DEEP-ENCODE and RFECS | Deep neural network (DNN) | Acc 91.6% |
Liu, B., Fang, L., Long, R., Lan, X., & Chou, K. C.15 iEnhancer-2L15 | Chromatin state information of nine cell lines, including H1ES, K562, GM12878, HepG2, HUVEC, HSMM, NHLF, NHEK and HMEC | Physical structural property of inucleotide (Rise P1, Roll P2, Shift P3, Slide P4, Tilt P5, Twist p6) | SVM classification with RBF kernel function | Acc 76.89%, AUC 0.85 |
Histone modification marks | Sequence characteristics | Ensemble SVM | Acc 90.2% | |
Rajagopal, N., Xie, W., Li, Y., Wagner, U., Wang, W., Stamatoyannopoulos, J., & Ren, B.12, RFECS12 | 24 Histone modifications in two distinct human cell types, embryonic stem cells and lung fibroblasts (H1 and IMR90 datasets) | p300 ENCODE data in H1 and made enhancer predictions in 12 ENCODE cell-types using the three marks H3K4me1, H3K4me3 and H3K27ac Multiple chromatin marks | Random forests | Acc 95% |
Histone epigenetic marks | Optimum combination of Epigenetic profiles computed at various window sizes (1, 2.5, 5, 7.5, 10, 12.5 and 15 kb). | Genetic algorithm optimized Support vector machines | Acc 85.1% AUC 0.966 | |
Proposed method | VISTA Enhancer Browser (experimentally validated hg19) | K-mer frequency, Statistical and Non-linear features (sd, dfa, hurst, sampan, ac, rvntsl, ac_200, ac_300) | Ensemble Method (Bagged Tree) | Acc 93.3%, AUC 0.91 on test data from VISTA Enhancer Browser |
