Sequencing facility and DNA source associated patterns of virus-mappable reads in whole-genome sequencing data

VirusPredictor: XGBoost-based software to predict virus-related sequences in human data

MOTIVATION

Discovering disease causative pathogens, particularly viruses without reference genomes, poses a technical challenge as they are often unidentifiable through sequence alignment. Machine learning prediction of patient high-throughput sequences unmappable to human and pathogen genomes may reveal sequences originating from uncharacterized viruses. Currently, there is a lack of software specifically designed for accurately predicting such viral sequences in human data.

RESULTS

We developed a fast XGBoost method and software VirusPredictor leveraging an in-house viral genome database. Our two-step XGBoost models first classify each query sequence into one of three groups: infectious virus, endogenous retrovirus (ERV) or non-ERV human. The prediction accuracies increased as the sequences became longer, i.e. 0.76, 0.93, and 0.98 for 150-350 (Illumina short reads), 850-950 (Sanger sequencing data), and 2000-5000 bp sequences, respectively. Then, sequences predicted to be from infectious viruses are further classified into one of six virus taxonomic subgroups, and the accuracies increased from 0.92 to >0.98 when query sequences increased from 150-350 to >850 bp. The results suggest that Illumina short reads should be de novo assembled into contigs (e.g. ∼1000 bp or longer) before prediction whenever possible. We applied VirusPredictor to multiple real genomic and metagenomic datasets and obtained high accuracies. VirusPredictor, a user-friendly open-source Python software, is useful for predicting the origins of patients' unmappable sequences. This study is the first to classify ERVs in infectious viral sequence prediction. This is also the first study combining virus sub-group predictions.

AVAILABILITY AND IMPLEMENTATION

www.dllab.org/software/VirusPredictor.html.

Metagenomic analysis of viral genes integrated in whole genome sequencing data of Thai patients with Brugada syndrome

Brugada syndrome (BS) is an autosomal dominant inheritance cardiac arrhythmia disorder associated with sudden death in young adults. Thailand has the highest prevalence of BS worldwide, and over 60% of patients with BS still have unclear disease etiology. Here, we performeda new viral metagenome analysis pipeline called VIRIN and validated it with whole genome sequencing (WGS) data of HeLa cell lines and hepatocellular carcinoma. Then the VIRIN pipelinewas applied to identify viral integration positions from unmapped WGS data of Thai males, including 100 BS patients (case) and 100 controls. Even though the sample preparation had noviral enrichment step, we can identify several virus genes from our analysis pipeline. The predominance of human endogenous retrovirus K (HERV-K) viruses was found in both cases andcontrols by blastn and blastx analysis. This study is the first report on the full-length HERV-Kassembled genomes in the Thai population. Furthermore, the HERV-K integration breakpointpositions were validated and compared between the case and control datasets. Interestingly,Brugada cases contained HERV-K integration breakpoints at promoters five times more oftenthan controls. Overall, the highlight of this study is the BS-specific HERV-K breakpoint positionsthat were found at the gene coding region "NBPF11" (n = 9), "NBPF12" (n = 8) and longnon-coding RNA (lncRNA) "PCAT14" (n = 4) region. The genes and the lncRNA have been reported to be associated with congenital heart and arterial diseases. These findings provide another aspect of the BS etiology associated with viral genome integrations within the humangenome.

Characterization of Hepatitis B Virus Integrations Identified in Hepatocellular Carcinoma Genomes

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related mortality. Almost half of HCC cases are associated with hepatitis B virus (HBV) infections, which often lead to HBV sequence integrations in the human genome. Accurate identification of HBV integration sites at a single nucleotide resolution is critical for developing a better understanding of the cancer genome landscape and of the disease itself. Here, we performed further analyses and characterization of HBV integrations identified by our recently reported VIcaller platform in recurrent or known HCC genes (such as TERT, MLL4, and CCNE1) as well as non-recurrent cancer-related genes (such as CSMD2, NKD2, and RHOU). Our pathway enrichment analysis revealed multiple pathways involving the alcohol dehydrogenase 4 gene, such as the metabolism pathways of retinol, tyrosine, and fatty acid. Further analysis of the HBV integration sites revealed distinct patterns involving the integration upper breakpoints, integrated genome lengths, and integration allele fractions between tumor and normal tissues. Our analysis also implies that the VIcaller method has diagnostic potential through discovering novel clonal integrations in cancer-related genes. In conclusion, although VIcaller is a hypothesis free virome-wide approach, it can still be applied to accurately identify genome-wide integration events of a specific candidate virus and their integration allele fractions.