Viral Sequence Identification in Metagenomes using Natural Language Processing Techniques

Abstract

A B S T R A C TViral reads identification is one of the important steps in metagenomic data analysis. It shows up the diversity of the microbial communities and the functional characteristics of microorganisms. There are various tools that can identify viral reads in mixed metagenomic data using similarity and statistical tools. However, the lack of available genome diversity is a serious limitation to the existing techniques. In this work, we applied natural language processing approaches for document classification in analyzing metagenomic sequences. Text featurization is presented by treating DNA similar to natural language. These techniques reveal the importance of using the text feature extraction pipeline in sequence identification by transforming DNA base pairs into a set of characters with a term frequency and inverse document frequency techniques. Various machine learning classification algorithms are applied to viral identification tasks such as logistic regression and multi-layer perceptron. Moreover, we compared classical machine learning algorithms with VirFinder and VirNet, our deep attention model for viral reads identification on generated fragments of viruses and bacteria for benchmarking viral reads identification tools. Then, as a verification of our tool, It was applied to a simulated microbiome and virome data for tool verification and real metagenomic data of Roche 454 and Illumina for a case study.

Applying Deep Learning for Genome Detection of Coronavirus

“…Using this information, the tool recognizes the genomic sequence of the virus with an accuracy of 87.5%. They improved the accuracy reported in [12] by 0.5%. This tool is freely available online.…”

“…This can be a useful clue for the clinicians to find the most effective vaccine or drug for the treatment of 'COVID-19'. The comparison of the proposed CNN model 'GenomeSimilarityPredictor' with models proposed in the literature [10][11][12][13][14][15][16][17][18][19][20][21] shows that model has reported higher accuracy and outperforms the existing techniques as shown in Fig 8. Its effectiveness in dealing with noisy data, low time complexity makes it applicable for the screening of infected genomes in the present situation of 'Global Pandemic'. The zero instance in the FP and only 1 instance in the FN increase the acceptability of this model.…”

“…They claimed that genomic sequence detection is a fast and reliable technique for diagnosis of the disease. The researchers in [10][11][12][13][14][15][16][17][18][19][20][21] proposed the computer-based solutions for the detection of the viral genome or predicting the genomic similarity among viruses. The work proposed in [10] employs the hidden Markov's model and identifies the viral genome from the host cell with an accuracy of 87%.…”

“…But, its low accuracy is a point of concern for its use in the screening of patients in real-time. The authors in [12] applied natural language processing to detect the genome of a viral sequence. They considered the genome as a string and detected the sequence of base pairs as a sub-string.…”

Genome binning of viral entities from bulk metagenomics data

“…Currently, several approaches for identifying viral sequences in metagenomics data exist and have helped in supersizing viral databases of uncultivated viral genomes (UViGs) over the last few years [20][21][22] . These tools are often based on sequence similarity 23 , sequence composition [24][25][26][27]28,29 , and identification of viral proteins or the lack of cellular ones 28,29 . A common denominator for these tools is their per-contig/sequence virus evaluation approach that is not optimal for addressing fragmented multi-contig virus assemblies.…”