A comparison on predicting functional impact of genomic variants

ABSTRACT Single-nucleotide polymorphism (SNPs) may cause the diverse functional impact on RNA or protein changing genotype and phenotype, which may lead to common or complex diseases like cancers. Accurate prediction of the functional impact of SNPs is crucial to discover the ‘influential’ (deleterious, pathogenic, disease-causing, and predisposing) variants from massive background polymorphisms in the human genome. Increasing computational methods have been developed to predict the functional impact of variants. However, predictive performances of these computational methods on massive genomic variants are still unclear. In this regard, we systematically evaluated 14 important computational methods including specific methods for one type of variant and general methods for multiple types of variants from several aspects; none of these methods achieved excellent (AUC ≥ 0.9) performance in both data sets. CADD and REVEL achieved excellent performance on multiple types of variants and missense variants, respectively. This comparison aims to assist researchers and clinicians to select appropriate methods or develop better predictive methods.

A robust benchmark for evaluating and improving mosaic variant calling strategies

The rapid advances in sequencing and analysis technologies have enabled the accurate detection of diverse forms of genomic variants, including germline, somatic, and mosaic mutations. However, unlike for the former two mutations, the best practices for mosaic variant calling still remain chaotic due to the technical and conceptual difficulties faced in evaluation. Here, we present our benchmark of nine feasible strategies for mosaic variant detection based on a systematically designed reference standard that mimics mosaic samples, with 390,153 control positive and 35,208,888 negative single-nucleotide variants and insertion–deletion mutations. We identified the condition-dependent strengths and weaknesses of the current strategies, instead of a single winner, regarding variant allele frequencies, variant sharing, and the usage of control samples. Moreover, feature-level investigation directs the way for immediate to prolonged improvements in mosaic variant calling. Our results will guide researchers in selecting suitable calling algorithms and suggest future strategies for developers.

A modified decision tree approach to improve the prediction and mutation discovery for drug resistance in Mycobacterium tuberculosis

Background Drug resistant Mycobacterium tuberculosis is complicating the effective treatment and control of tuberculosis disease (TB). With the adoption of whole genome sequencing as a diagnostic tool, machine learning approaches are being employed to predict M. tuberculosis resistance and identify underlying genetic mutations. However, machine learning approaches can overfit and fail to identify causal mutations if they are applied out of the box and not adapted to the disease-specific context. We introduce a machine learning approach that is customized to the TB setting, which extracts a library of genomic variants re-occurring across individual studies to improve genotypic profiling. Results We developed a customized decision tree approach, called Treesist-TB, that performs TB drug resistance prediction by extracting and evaluating genomic variants across multiple studies. The application of Treesist-TB to rifampicin (RIF), isoniazid (INH) and ethambutol (EMB) drugs, for which resistance mutations are known, demonstrated a level of predictive accuracy similar to the widely used TB-Profiler tool (Treesist-TB vs. TB-Profiler tool: RIF 97.5% vs. 97.6%; INH 96.8% vs. 96.5%; EMB 96.8% vs. 95.8%). Application of Treesist-TB to less understood second-line drugs of interest, ethionamide (ETH), cycloserine (CYS) and para-aminosalisylic acid (PAS), led to the identification of new variants (52, 6 and 11, respectively), with a high number absent from the TB-Profiler library (45, 4, and 6, respectively). Thereby, Treesist-TB had improved predictive sensitivity (Treesist-TB vs. TB-Profiler tool: PAS 64.3% vs. 38.8%; CYS 45.3% vs. 30.7%; ETH 72.1% vs. 71.1%). Conclusion Our work reinforces the utility of machine learning for drug resistance prediction, while highlighting the need to customize approaches to the disease-specific context. Through applying a modified decision learning approach (Treesist-TB) across a range of anti-TB drugs, we identified plausible resistance-encoding genomic variants with high predictive ability, whilst potentially overcoming the overfitting challenges that can affect standard machine learning applications.

Monogenic forms of DSD: An update

DSD encompasses a wide range of pathologies that impact gonad formation, development and function in both 46,XX and 46,XY individuals. The majority of these conditions are considered to be monogenic, although the expression of the phenotype may be influenced by genetic modifiers. Although considered monogenic, establishing the genetic etiology in DSD has been difficult compared to other congenital disorders for a number of reasons including the absence of family cases for classical genetic association studies and the lack of evolutionary conservation of key genetic factors involved in gonad formation. In recent years, the widespread use of genomic sequencing technologies has resulted in multiple genes being identified and proposed as novel monogenic causes of 46,XX and/or 46,XY DSD. In this review, we will focus on the main genomic findings of recent years, which consists of new candidate genes or loci for DSD as well as new reproductive phenotypes associated with genes that are well established to cause DSD. For each gene or loci, we summarise the data that is currently available in favor of or against a role for these genes in DSD or the contribution of genomic variants within well-established genes to a new reproductive phenotype. Based on this analysis we propose a series of recommendations that should aid the interpretation of genomic data and ultimately help to improve the accuracy and yield genetic diagnosis of DSD.

Novel SNPs and haplotypes identified in the CD4 gene and their influence on deregressed MACE EBV indexes of milk-related traits in Simmental breed

Cluster of differentiation 4 (CD4) is the accessory protein non-covalently bound to the T cell receptor that recognizes an invariant region of MHC class II on antigen presenting cells. Its cytoplasmic tail, physically associated with a protein tyrosine kinase, is important in the activation of helper/inducer T lymphocytes. In Bos taurus, CD4 gene is located on chromosome 5 from which two isoforms are transcribed, with a different number of amino acids due to splicing of exon 7 and variation in the reading frame. The aim of this study was to investigate the sequence of the entire CD4 gene in Simmental sires to evaluate the effects of genomic variants on the indexes of the bulls for milk, fat and protein yields, as well as somatic cell score. The associations among genomic variants and indexes were analysed using the Allele and GLM procedures of SAS 9.4. The analysis indicated that only four of the thirty-one identified SNPs influenced the considered traits. Identified variants insist on coding zones and intronic sequences, where we revealed the presence of sites for transcription factors. To evaluate the existence of haplotypic effects, combinations among the four genomic variants (SNP 3, SNP 8, SNP 11 and SNP 19) were investigated. Six different haplotypic alleles were identified, but only four of them were frequent enough to allow for an evaluation of any haplotypic effect (at least six copies in the examined sample): Hap1, Hap2, Hap3 and Hap6. The analysis of associations between the selected haplotypes in the CD4 gene with milk related indexes showed that bulls with Hap2 (T-A-C-C) had better indexes for milk and protein yields (P < 0.05), whereas the presence of the Hap1 haplotype (A-G-A-T) caused a significant decrease of the index for protein yield (P < 0.05). Frequencies of the two alleles Hap1 and Hap2 (9 and 36% respectively) make them of interest for their possible inclusion in breeding schemes and support the hypothesis of considering this gene as a candidate for the improvement of milk-related traits in the Simmental breed.

Genome sequencing reveals the role of rare genomic variants in Chinese patients with symptomatic intracranial atherosclerotic disease

ObjectivesThe predisposition of intracranial atherosclerotic disease (ICAD) to East Asians over Caucasians infers a genetic basis which, however, remains largely unknown. Higher prevalence of vascular risk factors (VRFs) in Chinese over Caucasian patients who had a stroke, and shared risk factors of ICAD with other stroke subtypes indicate genes related to VRFs and/or other stroke subtypes may also contribute to ICAD.MethodsUnrelated symptomatic patients with ICAD were recruited for genome sequencing (GS, 60-fold). Rare and potentially deleterious single-nucleotide variants (SNVs) and small insertions/deletions (InDels) were detected in genome-wide and correlated to genes related to VRFs and/or other stroke subtypes. Rare aneuploidies, copy number variants (CNVs) and chromosomal structural rearrangements were also investigated. Lastly, candidate genes were used for pathway and gene ontology enrichment analysis.ResultsAmong 92 patients (mean age at stroke onset 61.0±9.3 years), GS identified likely ICAD-associated rare genomic variants in 54.3% (50/92) of patients. Forty-eight patients (52.2%, 48/92) had 59 rare SNVs/InDels reported or predicted to be deleterious in genes related to VRFs and/or other stroke subtypes. None of the 59 rare variants were identified in local subjects without ICAD (n=126). 31 SNVs/InDels were related to conventional VRFs, and 28 were discovered in genes related to other stroke subtypes. Our study also showed that rare CNVs (n=7) and structural rearrangement (a balanced translocation) were potentially related to ICAD in 8.7% (8/92) of patients. Lastly, candidate genes were significantly enriched in pathways related to lipoprotein metabolism and cellular lipid catabolic process.ConclusionsOur GS study suggests a role of rare genomic variants with various variant types contributing to the development of ICAD in Chinese patients.

Genomic Variants and Multilevel Regulation of ABCA1, ABCG1, and SCARB1 Expression in Atherogenesis

Atheroprotective properties of human plasma high-density lipoproteins (HDLs) are determined by their involvement in reverse cholesterol transport (RCT) from the macrophage to the liver. ABCA1, ABCG1, and SR-BI cholesterol transporters are involved in cholesterol efflux from macrophages to lipid-free ApoA-I and HDL as a first RCT step. Molecular determinants of RCT efficiency that may possess diagnostic and therapeutic meaning remain largely unknown. This review summarizes the progress in studying the genomic variants of ABCA1, ABCG1, and SCARB1, and the regulation of their function at transcriptional and post-transcriptional levels in atherosclerosis. Defects in the structure and function of ABCA1, ABCG1, and SR-BI are caused by changes in the gene sequence, such as single nucleotide polymorphism or various mutations. In the transcription initiation of transporter genes, in addition to transcription factors, long noncoding RNA (lncRNA), transcription activators, and repressors are also involved. Furthermore, transcription is substantially influenced by the methylation of gene promoter regions. Post-transcriptional regulation involves microRNAs and lncRNAs, including circular RNAs. The potential biomarkers and targets for atheroprotection, based on molecular mechanisms of expression regulation for three transporter genes, are also discussed in this review.

miR1908-5p regulates energy homeostasis in hepatocyte models

We previously identified genomic variants that are quantitative trait loci for circulating miR-1908-5p and then showed this microRNA to causally associate with plasma levels of LDL-C, fasting blood glucose and HbA1c. The link to LDL-C was subsequently validated and clarified by the identification of a miR1908-5p-TGFB-LDLR regulatory axis. Here, we continue our investigations on miR1908-5p function by leveraging human primary hepatocytes and HuH-7 hepatoma models. Expression of miR1908-5p was shown to be sensitive to glucose and agents affecting glucose metabolism. Transcriptome-wide changes in primary hepatocytes and HuH-7 cells treated with a miR1908-5p mimic were investigated by enrichment approaches to identify targeted transcripts and cognate pathways. Significant pathways included autophagy and increased mitochondrial function. Reduced activation and/or levels of several key energy and metabolic regulators (AKT, mTOR, ME1, G6PD, AMPK and LKB) were subsequently confirmed in mimic treated HuH-7 cells. These effects were associated with reduced NADPH to NADP+ ratio in HuH-7 cells. LKB1 was validated as a direct target of miR1908-5p, the reintroduction of which was however insufficient to compensate for the impact of the miR1908-5p mimic on AMPK and ACC1. These findings implicate miR1908-5p in metabolic and energy regulation in hepatocyte models via multiple, independent, pathways.