DeeplyEssential: a deep neural network for predicting essential genes in microbes

Abstract

Background Essential genes are those genes that are critical for the survival of an organism. The prediction of essential genes in bacteria can provide targets for the design of novel antibiotic compounds or antimicrobial strategies. Results We propose a deep neural network for predicting essential genes in microbes. Our architecture called DeeplyEssential makes minimal assumptions about the input data (i.e., it only uses gene primary sequence and the corresponding protein sequence) to carry out the prediction thus maximizing its practical application compared to existing predictors that require structural or topological features which might not be readily available. We also expose and study a hidden performance bias that effected previous classifiers. Extensive results show that DeeplyEssential outperform existing classifiers that either employ down-sampling to balance the training set or use clustering to exclude multiple copies of orthologous genes. Conclusion Deep neural network architectures can efficiently predict whether a microbial gene is essential (or not) using only its sequence information.

Essential gene prediction using limited gene essentiality information–An integrative semi-supervised machine learning strategy

“…Based on the availability of labeled data of essential genes, researchers have employed supervised machine learning strategies [6][7][8] as well as deep learning-based strategies to predict essential genes [24,25]. The key advantage of these strategies lies in the fact that these models are capable of capturing the inherent patterns of a large array of biologically relevant 'features' that are distinctive and reflect the heterogeneous properties of essential genes.…”

“…Supervised machine learning classifiers such as logistic regression [ 26 , 27 ], support vector machine [ 28 – 31 ], random forest [ 32 ], decision tree [ 26 ], ensemble [ 26 ] and probabilistic Bayesian-based methods [ 26 , 27 , 33 ] and instance-based learning methods such as K Nearest neighbor (K-NN) and Weighted KNN (WKNN) [ 34 ] have been used for gene essentiality prediction. Deep Learning strategies based on multilayer perceptron networks have also been used for essential gene prediction [ 24 , 35 ]. In these studies, researchers have mostly opted for simpler optimization methods for parameter tuning, such as the grid search technique, where the entire parameter space is explored in all possible combinations.…”

Selected Research Articles from the 2019 International Workshop on Computational Network Biology: Modeling, Analysis, and Control (CNB-MAC)

“…After the workshop, ten original research papers [1][2][3][4][5][6][7][8][9][10] were accepted for publication in the CNB-MAC 2019 partner journals: BMC Bioinformatics and BMC Genomics. In the following we provide a brief summary of these selected papers.…”

“…The identification of essential genes in bacteria not only allows life scientists to determine the set of genes that are critical for the survival of an organism, it can also provide targets for antimicrobial/antibiotic drugs and the creation of self-sustaining artificial genomes. DeeplyEssential [6] leverages a deep neural network architecture for the identification of bacterial essential genes exclusively from the primary DNA sequence, thus maximizing the practicality of the tool.…”

“…Byung-Jun Yoon, Xiaoning Qian, Tamer Kahveci, and Ranadip Pal. The workshop featured original research papers [1][2][3][4][5][6][7][8][9][10], a highlight presentation of a recently published journal paper [11], and poster presentations [12,13], which were selected by the workshop chairs based on the reviews performed by the technical committee members. Reports from previous CNB-MAC workshops are available at [14][15][16].…”