Assessing the reproducibility of exome copy number variations predictions

Unique Capabilities of Genome Sequencing for Rare Disease Diagnosis

Background

Causal variants underlying rare disorders may remain elusive even after expansive gene panels or exome sequencing (ES). Clinicians and researchers may then turn to genome sequencing (GS), though the added value of this technique and its optimal use remain poorly defined. We therefore investigated the advantages of GS within a phenotypically diverse cohort.

Methods

GS was performed for 744 individuals with rare disease who were genetically undiagnosed. Analysis included review of single nucleotide, indel, structural, and mitochondrial variants.

Results

We successfully solved 218/744 (29.3%) cases using GS, with most solves involving established disease genes (157/218, 72.0%). Of all solved cases, 148 (67.9%) had previously had non-diagnostic ES. We systematically evaluated the 218 causal variants for features requiring GS to identify and 61/218 (28.0%) met these criteria, representing 8.2% of the entire cohort. These included small structural variants (13), copy neutral inversions and complex rearrangements (8), tandem repeat expansions (6), deep intronic variants (15), and coding variants that may be more easily found using GS related to uniformity of coverage (19).

Conclusion

We describe the diagnostic yield of GS in a large and diverse cohort, illustrating several types of pathogenic variation eluding ES or other techniques. Our results reveal a higher diagnostic yield of GS, supporting the utility of a genome-first approach, with consideration of GS as a secondary or tertiary test when higher-resolution structural variant analysis is needed or there is a strong clinical suspicion for a condition and prior targeted genetic testing has been negative.

Technological and computational approaches to detect somatic mosaicism in epilepsy

Lesional epilepsy is a common and severe disease commonly associated with malformations of cortical development, including focal cortical dysplasia and hemimegalencephaly. Recent advances in sequencing and variant calling technologies have identified several genetic causes, including both short/single nucleotide and structural somatic variation. In this review, we aim to provide a comprehensive overview of the methodological advancements in this field while highlighting the unresolved technological and computational challenges that persist, including ultra-low variant allele fractions in bulk tissue, low availability of paired control samples, spatial variability of mutational burden within the lesion, and the issue of false-positive calls and validation procedures. Information from genetic testing in focal epilepsy may be integrated into clinical care to inform histopathological diagnosis, postoperative prognosis, and candidate precision therapies.

Exome sequencing as a first-tier test for copy number variant detection: retrospective evaluation and prospective screening in 2418 cases

BACKGROUND

Despite the availability of whole exome (WES) and genome sequencing (WGS), chromosomal microarray (CMA) remains the first-line diagnostic test in most rare disorders diagnostic workup, looking for copy number variations (CNVs), with a diagnostic yield of 10%-20%. The question of the equivalence of CMA and WES in CNV calling is an organisational and economic question, especially when ordering a WGS after a negative CMA and/or WES.

METHODS

This study measures the equivalence between CMA and GATK4 exome sequencing depth of coverage method in detecting coding CNVs on a retrospective cohort of 615 unrelated individuals. A prospective detection of WES-CNV on a cohort of 2418 unrelated individuals, including the 615 individuals from the validation cohort, was performed.

RESULTS

On the retrospective validation cohort, every CNV detectable by the method (ie, a CNV with at least one exon not in a dark zone) was accurately called (64/64 events). In the prospective cohort, 32 diagnoses were performed among the 2418 individuals with CNVs ranging from 704 bp to aneuploidy. An incidental finding was reported. The overall increase in diagnostic yield was of 1.7%, varying from 1.2% in individuals with multiple congenital anomalies to 1.9% in individuals with chronic kidney failure.

CONCLUSION

Combining single-nucleotide variant (SNV) and CNV detection increases the suitability of exome sequencing as a first-tier diagnostic test for suspected rare Mendelian disorders. Before considering the prescription of a WGS after a negative WES, a careful reanalysis with updated CNV calling and SNV annotation should be considered.

Algorithmic improvements for discovery of germline copy number variants in next-generation sequencing data

Background

Copy number variants (CNVs) play a significant role in human heredity and disease. However, sensitive and specific characterization of germline CNVs from NGS data has remained challenging, particularly for hybridization-capture data in which read counts are the primary source of copy number information.

Results

We describe two algorithmic adaptations that improve CNV detection accuracy in a Hidden Markov Model (HMM) context. First, we present a method for computing target- and copy number-specific emission distributions. Second, we demonstrate that the Pointwise Maximum a posteriori (PMAP) HMM decoding procedure yields improved sensitivity for small CNV calls compared to the more common Viterbi HMM decoder. We develop a prototype implementation, called Cobalt, and compare it to other CNV detection tools using sets of simulated and previously detected CNVs with sizes spanning a single exon to a full chromosome.

Conclusions

In both the simulation and previously detected CNV studies Cobalt shows similar sensitivity but significantly fewer false positive detections compared to other callers. Overall sensitivity is 80–90% for deletion CNVs spanning 1–4 targets and 90–100% for larger deletion events, while sensitivity is somewhat lower for small duplication CNVs.

Comprehensive analysis of recessive carrier status using exome and genome sequencing data in 1543 Southern Chinese

Traditional carrier screening has been utilized for the detection of carriers of genetic disorders. Since a comprehensive assessment of the carrier frequencies of recessive conditions in the Southern Chinese population is not yet available, we performed a secondary analysis on the spectrum and carrier status for 315 genes causing autosomal recessive disorders in 1543 Southern Chinese individuals with next-generation sequencing data, 1116 with exome sequencing and 427 with genome sequencing data. Our data revealed that 1 in 2 people (47.8% of the population) was a carrier for one or more recessive conditions, and 1 in 12 individuals (8.30% of the population) was a carrier for treatable inherited conditions. In alignment with current American College of Obstetricians and Gynecologists (ACOG) pan-ethnic carrier recommendations, 1 in 26 individuals were identified as carriers of cystic fibrosis, thalassemia, and spinal muscular atrophy in the Southern Chinese population. When the >1% expanded carrier screening rate recommendation by ACOG was used, 11 diseases were found to meet the criteria in the Southern Chinese population. Approximately 1 in 3 individuals (35.5% of the population) were carriers of these 11 conditions. If the 1 in 200 carrier frequency threshold is used, and additional seven genes would meet the criteria, and 2 in 5 individuals (38.7% of the population) would be detected as a carrier. This study provides a comprehensive catalogue of the carrier spectrum and frequency in the Southern Chinese population and can serve as a reference for careful evaluation of the conditions to be included in expanded carrier screening for Southern Chinese people.

Diagnostic pitfalls in patients with malformations of cortical development

Malformations of cortical development (MCDs) are a major source of morbidity and mortality in the pediatric patient cohort. Correct diagnosis of the cause is essential for symptom management, disease prognosis and family counselling but is frequently hampered due to numerous potential pitfalls in the diagnostic process. This review highlights potential problems that either prevent the establishment of a diagnosis or are the sources of diagnostic errors. The focus is placed on hereditary causes of MCDs and strategies will be proposed to circumvent potential diagnostic pitfalls. Errors may occur during variant detection, filtering, or interpretation in relation to patient's phenotype. Based on detailed clinical assessment suitable targeted and untargeted methods to identify pathogenic variants with context-dependent filtering and evaluation approaches will be discussed.

Clinical exome sequencing - mistakes and caveats

Massive parallel sequencing technology has become the predominant technique for genetic diagnostics and research. Many genetic laboratories have wrestled with the challenges of setting up genetic testing workflows based on a completely new technology. The learning curve we went through as a laboratory was accompanied by growing pains while we gained new knowledge and expertise. Here we discuss some important mistakes that have been made in our laboratory through 10 years of clinical exome sequencing but that have given us important new insights on how to adapt our working methods. We provide these examples and the lessons that we learned to help other laboratories avoid to make the same mistakes.

Copy number variants calling from WES data through eXome hidden Markov model (XHMM) identifies additional 2.5% pathogenic genomic imbalances smaller than 30 kb undetected by array-CGH

It has been estimated that Copy Number Variants (CNVs) account for 10%-20% of patients affected by Developmental Disorder (DD)/Intellectual Disability (ID). Although array comparative genomic hybridization (array-CGH) represents the gold-standard for the detection of genomic imbalances, common Agilent array-CGH 4 × 180 kb arrays fail to detect CNVs smaller than 30 kb. Whole Exome sequencing (WES) is becoming the reference application for the detection of gene variants and makes it possible also to infer genomic imbalances at single exon resolution. However, the contribution of small CNVs in DD/ID is still underinvestigated. We made use of the eXome Hidden Markov Model (XHMM) software, a tool utilized by the ExAC consortium, to detect CNVs from whole exome sequencing data, in a cohort of 200 unsolved DD/DI patients after array-CGH and WES-based single nucleotide/indel variant analyses. In five out of 200 patients (2.5%), we identified pathogenic CNV(s) smaller than 30 kb, ranging from one to six exons. They included two heterozygous deletions in TCF4 and STXBP1 and three homozygous deletions in PPT1, CLCN2, and PIGN. After reverse phenotyping, all variants were reported as causative. This study shows the interest in applying sequencing-based CNV detection, from available WES data, to reduce the diagnostic odyssey of additional patients unsolved DD/DI patients and compare the CNV-detection yield of Agilent array-CGH 4 × 180kb versus whole exome sequencing.

VarGenius-HZD Allows Accurate Detection of Rare Homozygous or Hemizygous Deletions in Targeted Sequencing Leveraging Breadth of Coverage

Homozygous deletions (HDs) may be the cause of rare diseases and cancer, and their discovery in targeted sequencing is a challenging task. Different tools have been developed to disentangle HD discovery but a sensitive caller is still lacking. We present VarGenius-HZD, a sensitive and scalable algorithm that leverages breadth-of-coverage for the detection of rare homozygous and hemizygous single-exon deletions (HDs). To assess its effectiveness, we detected both real and synthetic rare HDs in fifty exomes from the 1000 Genomes Project obtaining higher sensitivity in comparison with state-of-the-art algorithms that each missed at least one event. We then applied our tool on targeted sequencing data from patients with Inherited Retinal Dystrophies and solved five cases that still lacked a genetic diagnosis. We provide VarGenius-HZD either stand-alone or integrated within our recently developed software, enabling the automated selection of samples using the internal database. Hence, it could be extremely useful for both diagnostic and research purposes.

Immune pathways and TP53 missense mutations are associated with longer survival in canine osteosarcoma

Osteosarcoma affects about 2.8% of dogs with cancer, with a one-year survival rate of approximately 45%. The purpose of this study was to characterize mutation and expression profiles of osteosarcoma and its association with outcome in dogs. The number of somatic variants identified across 26 samples ranged from 145 to 2,697 with top recurrent mutations observed in TP53 and SETD2. Additionally, 47 cancer genes were identified with copy number variations. Missense TP53 mutation status and low pre-treatment blood monocyte counts were associated with a longer disease-free interval (DFI). Patients with longer DFI also showed increased transcript levels of anti-tumor immune response genes. Although, T-cell and myeloid cell quantifications were not significantly associated with outcome; immune related genes, PDL-1 and CD160, were correlated with T-cell abundance. Overall, the association of gene expression and mutation profiles to outcome provides insights into pathogenesis and therapeutic interventions in osteosarcoma patients.

SILO: A Computational Method for Detecting Copy Number Gain in Clinical Specimens Analyzed on a Next-Generation Sequencing Platform

Next-generation sequencing (NGS) has proved to be a beneficial approach for genotyping solid tumor specimens and for identifying clinically actionable mutations. However, copy number variations (CNVs), which can be equally important, are often challenging to detect from NGS data. Current bioinformatics methods for CNV detection from NGS often require comparison of tumor/normal pairs and/or the sequencing of whole genome or whole exome. These approaches are currently impractical for routine clinical practice. However, clinical practice does involve repeated use of the same gene panel on a large number of specimens over a long period of time. We take advantage of this repetitiveness and present SILO: a procedure for CNV detection based on NGS on a gene panel. The SILO algorithm analyzes coverage depth of the aligned reads from a sample and predicts CNV by comparing this depth to the average depth seen in a large training set of other samples. Such comparison is robust and can reliably detect copy number gain, although it is found to be unreliable in detecting copy number losses. Successful validation of SILO on NGS data from the Ion Torrent platform with two panels is presented: a small hotspot panel and a larger cancer gene panel.

Incorporating Machine Learning into Established Bioinformatics Frameworks

The exponential growth of biomedical data in recent years has urged the application of numerous machine learning techniques to address emerging problems in biology and clinical research. By enabling the automatic feature extraction, selection, and generation of predictive models, these methods can be used to efficiently study complex biological systems. Machine learning techniques are frequently integrated with bioinformatic methods, as well as curated databases and biological networks, to enhance training and validation, identify the best interpretable features, and enable feature and model investigation. Here, we review recently developed methods that incorporate machine learning within the same framework with techniques from molecular evolution, protein structure analysis, systems biology, and disease genomics. We outline the challenges posed for machine learning, and, in particular, deep learning in biomedicine, and suggest unique opportunities for machine learning techniques integrated with established bioinformatics approaches to overcome some of these challenges.

Utility of droplet digital PCR and NGS-based CNV clinical assays in hearing loss diagnostics: current status and future prospects

Introduction: Genetic variants in over 100 genes can cause non-syndromic hearing loss (NSHL). Comprehensive diagnostic testing of these genes requires detecting pathogenic sequence and copy number alterations with economical, scalable and sensitive assays. Here we discuss best practices and effective testing algorithms for hearing-loss-related genes with special emphasis on detection of copy number variants.Areas covered: We review studies that used next-generation sequencing (NGS), chromosomal microarrays, droplet digital PCR (ddPCR), and multiplex ligation-dependent probe amplification (MLPA) for the diagnosis of NSHL. We specifically focus on unique and recurrent copy number changes that affect the GJB2 and STRC genes, two of the most common causes of NSHL.Expert opinion: NGS panels and exome sequencing can detect most pathogenic sequence and copy number variants that cause NSHL; however, GJB2 and STRC currently require additional assays to capture all pathogenic copy number variants. Adoption of genome sequencing may simplify diagnostic workflows, but further investigational studies will be required to evaluate its clinical efficacy.

Performance of In Silico Prediction Tools for the Detection of Germline Copy Number Variations in Cancer Predisposition Genes in 4208 Female Index Patients with Familial Breast and Ovarian Cancer

Simple Summary

The identification of germline copy number variants (CNVs) by targeted nextgeneration sequencing frequently relies on in silico prediction tools with unknown sensitivities. We investigated the performances of four in silico CNV prediction tools in 17 cancer predisposition genes in a large series of 4208 female index patients with familial breast and/or ovarian cancer. We identified 77 CNVs in 76 out of 4208 patients; six CNVs were missed by at least one of the prediction tools. Experimental verification of in silico predicted CNVs is required due to high frequencies of false positive predictions. For female index patients with familial breast and/or ovarian cancer, CNV detection should not be restricted to BRCA1/2 due to the relevant proportion of CNVs in further cancer predisposition genes.

Abstract

The identification of germline copy number variants (CNVs) by targeted next-generation sequencing (NGS) frequently relies on in silico CNV prediction tools with unknown sensitivities. We investigated the performances of four in silico CNV prediction tools, including one commercial (Sophia Genetics DDM) and three non-commercial tools (ExomeDepth, GATK gCNV, panelcn.MOPS) in 17 cancer predisposition genes in 4208 female index patients with familial breast and/or ovarian cancer (BC/OC). CNV predictions were verified via multiplex ligation-dependent probe amplification. We identified 77 CNVs in 76 out of 4208 patients (1.81%); 33 CNVs were identified in genes other than BRCA1/2, mostly in ATM, CHEK2, and RAD51C and less frequently in BARD1, MLH1, MSH2, PALB2, PMS2, RAD51D, and TP53. The Sophia Genetics DDM software showed the highest sensitivity; six CNVs were missed by at least one of the non-commercial tools. The positive predictive values ranged from 5.9% (74/1249) for panelcn.MOPS to 79.1% (72/91) for ExomeDepth. Verification of in silico predicted CNVs is required due to high frequencies of false positive predictions, particularly affecting target regions at the extremes of the GC content or target length distributions. CNV detection should not be restricted to BRCA1/2 due to the relevant proportion of CNVs in further BC/OC predisposition genes.

inCNV: An Integrated Analysis Tool for Copy Number Variation on Whole Exome Sequencing

The detection of copy number variations (CNVs) on whole-exome sequencing (WES) represents a cost-effective technique for the study of genetic variants. This approach, however, has encountered an obstacle with high false-positive rates due to biases from exome sequencing capture kits and GC contents. Although plenty of CNV detection tools have been developed, they do not perform well with all types of CNVs. In addition, most tools lack features of genetic annotation, CNV visualization, and flexible installation, requiring users to put much effort into CNV interpretation. Here, we present “inCNV,” a web-based application that can accept multiple CNV-tool results, then integrate and prioritize them with user-friendly interfaces. This application helps users analyze the importance of called CNVs by generating CNV annotations from Ensembl, Database of Genomic Variants (DGV), ClinVar, and Online Mendelian Inheritance in Man (OMIM). Moreover, users can select and export CNVs of interest including their flanking sequences for primer design and experimental verification. We demonstrated how inCNV could help users filter and narrow down the called CNVs to a potentially novel CNV, a common CNV within a group of samples of the same disease, or a de novo CNV of a sample within the same family. Besides, we have provided in CNV as a docker image for ease of installation (https://github.com/saowwapark/inCNV).

A Novel, Recurrent, 3.6-kb Deletion in the PYGL Gene Contributes to Glycogen Storage Disease Type VI

The PYGL gene is the only established gene known to cause glycogen storage disease type VI (GSD6), which is a rare autosomal recessive disorder associated with hepatomegaly, elevated levels of hepatic transaminases, and hypoglycemia. Extended bioinformatics analysis was performed on the exome sequencing data of 5 patients who were clinically diagnosed as having or highly suspected of having GSD, and a single heterozygous pathogenic or likely pathogenic or rare variant of uncertain significance single-nucleotide variant was identified on the PYGL gene. A recurrent, novel, 3.6-kb deletion involving exons 14 to 17 of PYGL was identified in three of the five patients. Together with the two novel and one established stop-gain SNVs, they were diagnosed as compounds heterozygous of PYGL variants and confirmed as GSD6. The detected 3.6-kb deletion was further screened in a Chinese cohort of 31,317 individuals without hepatic abnormalities, and 10 carriers were identified, showing an allele frequency of 0.016%. Compared with the previously established 47 PYGL pathogenic or likely pathogenic SNVs, the novel pathogenic deletion had the second highest allele frequency among the population. This recurrent, novel, 3.6-kb deletion improved the molecular diagnostic rate of the GSD6. The relatively high frequency of the variant suggests that it is a potential mutation hotspot in patients with GSD6.

Evaluation of CNV detection tools for NGS panel data in genetic diagnostics

Although germline copy-number variants (CNVs) are the genetic cause of multiple hereditary diseases, detecting them from targeted next-generation sequencing data (NGS) remains a challenge. Existing tools perform well for large CNVs but struggle with single and multi-exon alterations. The aim of this work is to evaluate CNV calling tools working on gene panel NGS data and their suitability as a screening step before orthogonal confirmation in genetic diagnostics strategies. Five tools (DECoN, CoNVaDING, panelcn.MOPS, ExomeDepth, and CODEX2) were tested against four genetic diagnostics datasets (two in-house and two external) for a total of 495 samples with 231 single and multi-exon validated CNVs. The evaluation was performed using the default and sensitivity-optimized parameters. Results showed that most tools were highly sensitive and specific, but the performance was dataset dependant. When evaluating them in our diagnostics scenario, DECoN and panelcn.MOPS detected all CNVs with the exception of one mosaic CNV missed by DECoN. However, DECoN outperformed panelcn.MOPS specificity achieving values greater than 0.90 when using the optimized parameters. In our in-house datasets, DECoN and panelcn.MOPS showed the highest performance for CNV screening before orthogonal confirmation. Benchmarking and optimization code is freely available at https://github.com/TranslationalBioinformaticsIGTP/CNVbenchmarkeR .

Application of Whole-Exome Sequencing in Detecting Copy Number Variants in Patients with Developmental Delay and/or Multiple Congenital Malformations

Overcoming challenges for the unambiguous detection of copy number variations is essential to broaden our understanding of the role of genomic variants in the clinical phenotype. With the improvement of software and databases, whole-exome sequencing quickly can become an excellent strategy in the routine diagnosis of patients with a developmental delay and/or multiple congenital malformations. However, even after a detailed analysis of pathogenic single-nucleotide variants and indels in known disease genes, using whole-exome sequencing, some patients with suspected syndromic conditions are left without a conclusive diagnosis. These negative results could be the result of different factors including nongenetic etiologies, lack of knowledge about the genes that cause different disease phenotypes, or, in some cases, a deletion or duplication of genomic information not routinely detectable by whole-exome sequencing variant calling. Although copy number variant detection is possible using whole-exome sequencing data, such analysis presents significant challenges and cannot yet be used to replace chromosomal arrays for identification of deletions or duplications.

Clinical exome sequencing as the first-tier test for diagnosing developmental disorders covering both CNV and SNV: a Chinese cohort

Background

Developmental disorders (DDs) are early onset disorders affecting 5%–10% of children worldwide. Chromosomal microarray analysis detecting CNVs is currently recommended as the first-tier test for DD diagnosis. However, this analysis omits a high percentage of disease-causing single nucleotide variations (SNVs) that warrant further sequencing. Currently, next-generation sequencing can be used in clinical scenarios detecting CNVs, and the use of exome sequencing in the DD cohort ahead of the microarray test has not been evaluated.

Methods

Clinical exome sequencing (CES) was performed on 1090 unrelated Chinese DD patients who were classified into five phenotype subgroups. CNVs and SNVs were both detected and analysed based on sequencing data.

Results

An overall diagnostic rate of 41.38% was achieved with the combinational analysis of CNV and SNV. Over 12.02% of patients were diagnosed based on CNV, which was comparable with the published CMA diagnostic rate, while 0.74% were traditionally elusive cases who had dual diagnosis or apparently homozygous mutations that were clarified. The diagnostic rates among subgroups ranged from 21.82% to 50.32%. The top three recurrent cytobands with diagnostic CNVs were 15q11.2-q13.1, 22q11.21 and 7q11.23. The top three genes with diagnostic SNVs were: MECP2, SCN1A and SCN2A. Both the diagnostic rate and spectrums of CNVs and SNVs showed differences among the phenotype subgroups.

Conclusion

With a higher diagnostic rate, more comprehensive observation of variations and lower cost compared with conventional strategies, simultaneous analysis of CNVs and SNVs based on CES showed potential as a new first-tier choice to diagnose DD.

A highly sensitive and specific workflow for detecting rare copy-number variants from exome sequencing data

Background

Exome sequencing (ES) is a first-tier diagnostic test for many suspected Mendelian disorders. While it is routine to detect small sequence variants, it is not a standard practice in clinical settings to detect germline copy-number variants (CNVs) from ES data due to several reasons relating to performance. In this work, we comprehensively characterized one of the most sensitive ES-based CNV tools, ExomeDepth, against SNP array, a standard of care test in clinical settings to detect genome-wide CNVs.

Methods

We propose a modified ExomeDepth workflow by excluding exons with low mappability prior to variant calling to drastically reduce the false positives originating from the repetitive regions of the genome, and an iterative variant calling framework to assess the reproducibility. We used a cohort of 307 individuals with clinical ES data and clinical SNP array to estimate the sensitivity and false discovery rate of the CNV detection using exome sequencing. Further, we performed targeted testing of the STRC gene in 1972 individuals. To reduce the number of variants for downstream analysis, we performed a large-scale iterative variant calling process with random control cohorts to assess the reproducibility of the CNVs.

Results

The modified workflow presented in this paper reduced the number of total variants identified by one third while retaining a higher sensitivity of 97% and resulted in an improved false discovery rate of 11.4% compared to the default ExomeDepth pipeline. The exclusion of exons with low mappability removes 4.5% of the exons, including a subset of exons (0.6%) in disease-associated genes which are intractable by short-read next-generation sequencing (NGS). Results from the reproducibility analysis showed that the clinically reported variants were reproducible 100% of the time and that the modified workflow can be used to rank variants from high to low confidence. Targeted testing of 30 CNVs identified in STRC, a challenging gene to ascertain by NGS, showed a 100% validation rate.

Conclusions

In summary, we introduced a modification to the default ExomeDepth workflow to reduce the false positives originating from the repetitive regions of the genome, created a large-scale iterative variant calling framework for reproducibility, and provided recommendations for implementation in clinical settings.

Discordant and heterogeneous clinically relevant genomic alterations in circulating tumor cells vs plasma DNA from men with metastatic castration resistant prostate cancer

Circulating tumor cell (CTC) and cell-free (cf) DNA-based genomic alterations are increasingly being used for clinical decision-making in oncology. However, the concordance and discordance between paired CTC and cfDNA genomic profiles remain largely unknown. We performed comparative genomic hybridization (CGH) on CTCs and cfDNA, and low-pass whole genome sequencing (lpWGS) on cfDNA to characterize genomic alterations (CNA) and tumor content in two independent prospective studies of 93 men with mCRPC treated with enzalutamide/abiraterone, or radium-223. Comprehensive analysis of 69 patient CTCs and 72 cfDNA samples from 93 men with mCRPC, including 64 paired samples, identified common concordant gains in FOXA1, AR, and MYC, and losses in BRCA1, PTEN, and RB1 between CTCs and cfDNA. Concordant PTEN loss and discordant BRCA2 gain were associated with significantly worse outcomes in Epic AR-V7 negative men with mCRPC treated with abiraterone/enzalutamide. We identified and externally validated CTC-specific genomic alternations that were discordant in paired cfDNA, even in samples with high tumor content. These CTC/cfDNA-discordant regions included key genomic regulators of lineage plasticity, osteomimicry, and cellular differentiation, including MYCN gain in CTCs (31%) that was rarely detected in cfDNA. CTC MYCN gain was associated with poor clinical outcomes in AR-V7 negative men and small cell transformation. In conclusion, we demonstrated concordance of multiple genomic alterations across CTC and cfDNA platforms; however, some genomic alterations displayed substantial discordance between CTC DNA and cfDNA despite the use of identical copy number analysis methods, suggesting tumor heterogeneity and divergent evolution associated with poor clinical outcomes.

A machine-learning approach for accurate detection of copy number variants from exome sequencing

Copy number variants (CNVs) are a major cause of several genetic disorders, making their detection an essential component of genetic analysis pipelines. Current methods for detecting CNVs from exome-sequencing data are limited by high false-positive rates and low concordance because of inherent biases of individual algorithms. To overcome these issues, calls generated by two or more algorithms are often intersected using Venn diagram approaches to identify "high-confidence" CNVs. However, this approach is inadequate, because it misses potentially true calls that do not have consensus from multiple callers. Here, we present CN-Learn, a machine-learning framework that integrates calls from multiple CNV detection algorithms and learns to accurately identify true CNVs using caller-specific and genomic features from a small subset of validated CNVs. Using CNVs predicted by four exome-based CNV callers (CANOES, CODEX, XHMM, and CLAMMS) from 503 samples, we demonstrate that CN-Learn identifies true CNVs at higher precision (∼90%) and recall (∼85%) rates while maintaining robust performance even when trained with minimal data (∼30 samples). CN-Learn recovers twice as many CNVs compared to individual callers or Venn diagram-based approaches, with features such as exome capture probe count, caller concordance, and GC content providing the most discriminatory power. In fact, ∼58% of all true CNVs recovered by CN-Learn were either singletons or calls that lacked support from at least one caller. Our study underscores the limitations of current approaches for CNV identification and provides an effective method that yields high-quality CNVs for application in clinical diagnostics.

Copy number variant and runs of homozygosity detection by microarrays enabled more precise molecular diagnoses in 11,020 clinical exome cases

Background

Exome sequencing (ES) has been successfully applied in clinical detection of single nucleotide variants (SNVs) and small indels. However, identification of copy number variants (CNVs) using ES data remains challenging. The purpose of this study is to understand the contribution of CNVs and copy neutral runs of homozygosity (ROH) in molecular diagnosis of patients referred for ES.

Methods

In a cohort of 11,020 consecutive ES patients, an Illumina SNP array analysis interrogating mostly coding SNPs was performed as a quality control (QC) measurement and for CNV/ROH detection. Among these patients, clinical chromosomal microarray analysis (CMA) was performed at Baylor Genetics (BG) on 3229 patients, either before, concurrently, or after ES. We retrospectively analyzed the findings from CMA and the QC array.

Results

The QC array can detect ~ 70% of pathogenic/likely pathogenic CNVs (PCNVs) detectable by CMA. Out of the 11,020 ES cases, the QC array identified PCNVs in 327 patients and uniparental disomy (UPD) disorder-related ROH in 10 patients. The overall PCNV/UPD detection rate was 5.9% in the 3229 ES patients who also had CMA at BG; PCNV/UPD detection rate was higher in concurrent ES and CMA than in ES with prior CMA (7.2% vs 4.6%). The PCNVs/UPD contributed to the molecular diagnoses in 17.4% (189/1089) of molecularly diagnosed ES cases with CMA and were estimated to contribute in 10.6% of all molecularly diagnosed ES cases. Dual diagnoses with both PCNVs and SNVs were detected in 38 patients. PCNVs affecting single recessive disorder genes in a compound heterozygous state with SNVs were detected in 4 patients, and homozygous deletions (mostly exonic deletions) were detected in 17 patients. A higher PCNV detection rate was observed for patients with syndromic phenotypes and/or cardiovascular abnormalities.

Conclusions

Our clinical genomics study demonstrates that detection of PCNV/UPD through the QC array or CMA increases ES diagnostic rate, provides more precise molecular diagnosis for dominant as well as recessive traits, and enables more complete genetic diagnoses in patients with dual or multiple molecular diagnoses. Concurrent ES and CMA using an array with exonic coverage for disease genes enables most effective detection of both CNVs and SNVs and therefore is recommended especially in time-sensitive clinical situations.

Electronic supplementary material

The online version of this article (10.1186/s13073-019-0639-5) contains supplementary material, which is available to authorized users.

Evaluation of the performance of copy number variant prediction tools for the detection of deletions from whole genome sequencing data

BACKGROUND

Whole genome sequencing (WGS) has increased in popularity and decreased in cost over the past decade, rendering this approach as a viable and sensitive method for variant detection. In addition to its utility for single nucleotide variant detection, WGS data has the potential to detect Copy Number Variants (CNV) to fine resolution. Many CNV detection software packages have been developed exploiting four main types of data: read pair, split read, read depth, and assembly based methods. The aim of this study was to evaluate the efficiency of each of these main approaches in detecting germline deletions.

METHODS

WGS data and high confidence deletion calls for the individual NA12878 from the Genome in a Bottle consortium were the benchmark dataset. The performance of BreakDancer, CNVnator, Delly, FermiKit, and Pindel was assessed by comparing the accuracy and sensitivity of each software package in detecting deletions exceeding 1 kb.

RESULTS

There was considerable variability in the outputs of the different WGS CNV detection programs. The best performance was seen from BreakDancer and Delly, with 92.6% and 96.7% sensitivity, respectively and 34.5% and 68.5% false discovery rate (FDR), respectively. In comparison, Pindel, CNVnator, and FermiKit were less effective with sensitivities of 69.1%, 66.0%, and 15.8%, respectively and FDR of 91.3%, 69.0%, and 31.7%, respectively. Concordance across software packages was poor, with only 27 of the total 612 benchmark deletions identified by all five methodologies.

CONCLUSIONS

The WGS based CNV detection tools evaluated show disparate performance in identifying deletions ≥1 kb, particularly those utilising different input data characteristics. Software that exploits read pair based data had the highest sensitivity, namely BreakDancer and Delly. BreakDancer also had the second lowest false discovery rate. Therefore, in this analysis read pair methods (BreakDancer in particular) were the best performing approaches for the identification of deletions ≥1 kb, balancing accuracy and sensitivity. There is potential for improvement in the detection algorithms, particularly for reducing FDR. This analysis has validated the utility of WGS based CNV detection software to reliably identify deletions, and these findings will be of use when choosing appropriate software for deletion detection, in both research and diagnostic medicine.

Ximmer: a system for improving accuracy and consistency of CNV calling from exome data

Background

While exome and targeted next-generation DNA sequencing are primarily used for detecting single nucleotide changes and small indels, detection of copy number variants (CNVs) can provide highly valuable additional information from the data. Although there are dozens of exome CNV detection methods available, these are often difficult to use, and accuracy varies unpredictably between and within datasets.

Findings

We present Ximmer, a tool that supports an end-to-end process for evaluating, tuning, and running analysis methods for detection of CNVs in germline samples. Ximmer includes a simulation framework, implementations of several commonly used CNV detection methods, and a visualization and curation tool that together enable interactive exploration and quality control of CNV results. Using Ximmer, we comprehensively evaluate CNV detection on four datasets using five different detection methods. We show that application of Ximmer can improve accuracy and aid in quality control of CNV detection results. In addition, Ximmer can be used to run analyses and explore CNV results in exome data.

Conclusions

Ximmer offers a comprehensive tool and method for applying and improving accuracy of CNV detection methods for exome data.

CoDE-seq, an augmented whole-exome sequencing, enables the accurate detection of CNVs and mutations in Mendelian obesity and intellectual disability

Objective

The molecular diagnosis of extreme forms of obesity, in which accurate detection of both copy number variations (CNVs) and point mutations, is crucial for an optimal care of the patients and genetic counseling for their families. Whole-exome sequencing (WES) has benefited considerably this molecular diagnosis, but its poor ability to detect CNVs remains a major limitation. We aimed to develop a method (CoDE-seq) enabling the accurate detection of both CNVs and point mutations in one step.

Methods

CoDE-seq is based on an augmented WES method, using probes distributed uniformly throughout the genome. CoDE-seq was validated in 40 patients for whom chromosomal DNA microarray was available. CNVs and mutations were assessed in 82 children/young adults with suspected Mendelian obesity and/or intellectual disability and in their parents when available (n_total = 145).

Results

CoDE-seq not only detected all of the 97 CNVs identified by chromosomal DNA microarrays but also found 84 additional CNVs, due to a better resolution. When compared to CoDE-seq and chromosomal DNA microarrays, WES failed to detect 37% and 14% of CNVs, respectively. In the 82 patients, a likely molecular diagnosis was achieved in >30% of the patients. Half of the genetic diagnoses were explained by CNVs while the other half by mutations.

Conclusions

CoDE-seq has proven cost-efficient and highly effective as it avoids the sequential genetic screening approaches currently used in clinical practice for the accurate detection of CNVs and point mutations.

•

Whole-exome sequencing (WES) poorly detects CNVs.

•

Whole-genome sequencing remains expensive and hard to handle in clinical practice.

•

CoDE-seq (based on an augmented WES protocol) accurately detect CNVs.

•

CoDE-seq is highly effective for the diagnosis of obesity & intellectual disability.

A Screening Approach to Identify Clinically Actionable Variants Causing Congenital Heart Disease in Exome Data

BACKGROUND

Congenital heart disease (CHD)-structural abnormalities of the heart that arise during embryonic development-is the most common inborn malformation, affecting ≤1% of the population. However, currently, only a minority of cases can be explained by genetic abnormalities. The goal of this study was to identify disease-causal genetic variants in 30 families affected by CHD.

METHODS

Whole-exome sequencing was performed with the DNA of multiple family members. We utilized a 2-tiered whole-exome variant screening and interpretation procedure. First, we manually curated a high-confidence list of 90 genes known to cause CHD in humans, identified predicted damaging variants in genes on this list, and rated their pathogenicity using American College of Medical Genetics and Genomics-Association for Molecular Pathology guidelines.

RESULTS

In 3 families (10%), we found pathogenic variants in known CHD genes TBX5, TFAP2B, and PTPN11, explaining the cardiac lesions. Second, exomes were comprehensively analyzed to identify additional predicted damaging variants that segregate with disease in CHD candidate genes. In 10 additional families (33%), likely disease-causal variants were uncovered in PBX1, CNOT1, ZFP36L2, TEK, USP34, UPF2, KDM5A, KMT2C, TIE1, TEAD2, and FLT4.

CONCLUSIONS

The pathogenesis of CHD could be explained using our high-confidence CHD gene list for variant filtering in a subset of cases. Furthermore, our unbiased screening procedure of family exomes implicates additional genes and variants in the pathogenesis of CHD, which suggest themselves for functional validation. This 2-tiered approach provides a means of (1) identifying clinically actionable variants and (2) identifying additional disease-causal genes, both of which are essential for improving the molecular diagnosis of CHD.

Genetics of cerebellar disorders

The approach to identifying a genetic cause in patients with cerebellar disorders relies on history, examination, consultation, and testing, combined with specialized expertise because they are rare and genetically diverse. Cerebellar disorders can be caused by a variety of DNA alterations including single-nucleotide changes, small insertions or deletions, larger copy number variants, and nucleotide repeat expansions, exhibiting autosomal-recessive, autosomal-dominant (inherited and de novo), X-linked, and mitochondrial modes of inheritance. Imaging findings and a variety of neurologic and nonneurologic clinical features can help direct genetic testing and choose the most appropriate strategy. Clinical and genetic diagnoses are complementary, each providing distinct information for the care of the patient. In this chapter, we provide an overview of inheritance modes for different cerebellar disorders and the variety of genetic testing and tools that are currently available to reach a genetic diagnosis, including conventional and next-generation sequencing, classic, molecular and virtual cytogenetics, testing for repeat expansions, and other techniques. Practical examples are presented in both the text and accompanying vignettes.

ERDS-exome: a Hybrid Approach for Copy Number Variant Detection from Whole-exome Sequencing Data

Copy number variants (CNVs) play important roles in human disease and evolution. With the rapid development of next-generation sequencing technologies, many tools have been developed for inferring CNVs based on whole-exome sequencing (WES) data. However, as a result of the sparse distribution of exons in the genome, the limitations of the WES technique, and the nature of high-level signal noises in WES data, the efficacy of these variants remains less than desirable. Thus, there is need for the development of an effective tool to achieve a considerable power in WES CNVs discovery. In the present study, we describe a novel method, Estimation by Read Depth (RD) with Single-nucleotide variants from exome sequencing data (ERDS-exome). ERDS-exome employs a hybrid normalization approach to normalize WES data and to incorporate RD and single-nucleotide variation information together as a hybrid signal into a paired hidden Markov model to infer CNVs from WES data. Based on systematic evaluations of real data from the 1000 Genomes Project using other state-of-the-art tools, we observed that ERDS-exome demonstrates higher sensitivity and provides comparable or even better specificity than other tools. ERDS-exome is publicly available at: https://erds-exome.github.io.

A Path to Implement Precision Child Health Cardiovascular Medicine

Congenital heart defects (CHDs) affect approximately 1% of live births and are a major source of childhood morbidity and mortality even in countries with advanced healthcare systems. Along with phenotypic heterogeneity, the underlying etiology of CHDs is multifactorial, involving genetic, epigenetic, and/or environmental contributors. Clear dissection of the underlying mechanism is a powerful step to establish individualized therapies. However, the majority of CHDs are yet to be clearly diagnosed for the underlying genetic and environmental factors, and even less with effective therapies. Although the survival rate for CHDs is steadily improving, there is still a significant unmet need for refining diagnostic precision and establishing targeted therapies to optimize life quality and to minimize future complications. In particular, proper identification of disease associated genetic variants in humans has been challenging, and this greatly impedes our ability to delineate gene–environment interactions that contribute to the pathogenesis of CHDs. Implementing a systematic multileveled approach can establish a continuum from phenotypic characterization in the clinic to molecular dissection using combined next-generation sequencing platforms and validation studies in suitable models at the bench. Key elements necessary to advance the field are: first, proper delineation of the phenotypic spectrum of CHDs; second, defining the molecular genotype/phenotype by combining whole-exome sequencing and transcriptome analysis; third, integration of phenotypic, genotypic, and molecular datasets to identify molecular network contributing to CHDs; fourth, generation of relevant disease models and multileveled experimental investigations. In order to achieve all these goals, access to high-quality biological specimens from well-defined patient cohorts is a crucial step. Therefore, establishing a CHD BioCore is an essential infrastructure and a critical step on the path toward precision child health cardiovascular medicine.

A rare variant in the FHL1 gene associated with X-linked recessive hypoparathyroidism

Isolated familial hypoparathyroidism is an extremely rare disorder, which to date has been linked to several loci including mutations in CASR, GCM2, and PTH, as well as a rare condition defined as X-linked recessive hypoparathyroidism, previously associated with a 1.5 Mb region on Xq26-q27. Here, we report a patient with hypocalcemia-induced seizures leading to the diagnosis of primary hypoparathyroidism. Mutations in CASR, GCM2, and PTH were ruled out, while whole exome sequencing of the family suggested FHL1, located on chromosome Xq26, as the most likely causative gene variant (FHL1, exon 4, c.C283T, p.R95W). Since FHL1 has not been linked to calcium regulation before, we provide evidence for its functional role in hypoparathyroidism by: (i) bioinformatics analysis coupling its action to known modulators of PTH function; (ii) observing strong expression of fhl1b in Corpuscles of Stannius, gland-like aggregates in zebrafish that function in calcium regulation similar to mammalian PTH; and (iii) implicating fhl1b and FHL1 as regulators of calcium homeostasis in zebrafish and human cells, respectively. Altogether, our data suggest that FHL1 is a novel regulator of calcium homeostasis and implicate it as the causative gene for X-linked recessive hypoparathyroidism.