dv-trio: a family-based variant calling pipeline using DeepVariant

Motivation In 2018, Google published an innovative variant caller, DeepVariant, which converts pileups of sequence reads into images and uses a deep neural network to identify single-nucleotide variants and small insertion/deletions from next-generation sequencing data. This approach outperforms existing state-of-the-art tools. However, DeepVariant was designed to call variants within a single sample. In disease sequencing studies, the ability to examine a family trio (father-mother-affected child) provides greater power for disease mutation discovery. Results To further improve DeepVariant’s variant calling accuracy in family-based sequencing studies, we have developed a family-based variant calling pipeline, dv-trio, which incorporates the trio information from the Mendelian genetic model into variant calling based on DeepVariant. Availability and implementation dv-trio is available via an open source BSD3 license at GitHub (https://github.com/VCCRI/dv-trio/). Contact [email protected] Supplementary information Supplementary data are available at Bioinformatics online.

Family Trio-Based Whole Genome Optical Mapping Identifies Candidate Structural Variations Predisposing Children to Acute Lymphoblastic Leukemia

Germline mutations account for a substantial proportion of childhood cancer and may critically affect disease characteristics, therapy efficacy, severity of treatment side effects and patient outcome. To date, only 8-10% of childhood cancer cases can be explained by germline mutations identified in known cancer predisposing genes. This is in part due to the technical limitation of next generation short read sequencing, which detects single nucleotide variants, small deletions/insertions or simple copy number variations, but is not a reliable tool to identify larger structural variations (SVs, >500 bp) which are frequent in the human genome and may impact on disease predisposition. Using whole genome optical mapping (WGOM) we aimed at identification of de novo and inherited germline SVs in a cohort of patients with clinically suspected cancer predisposition but without informative findings in short read sequencing analyses. After informed consent we performed family trio based short read (2x 100 bp) whole exome sequencing (WES) on a HiSeq2500 (Illumina) and collected clinical and demographic data for a cohort of >100 families with children affected by cancer who were treated in our hospital. About 25% of the patients either (1) had a family history indicative of cancer susceptibility, or (2) had accompanying clinical findings (e.g. developmental delay, congenital anomalies) or (3) experienced excessive toxicity during chemotherapy. From this subgroup we selected four patients with acute lymphoblastic leukemia whose sequencing data and routine genetic workup were not informative of a known cancer predisposing syndrome and employed family trio-based next generation WGOM on a Saphyr instrument equipped with Access software (Bionano Genomics) to identify genomic SVs. To this end, we extracted and labeled high molecular weight DNA molecules at specific hexamer sequence motifs (average distance: 5 kb) using a DNA methyltransferase-based direct labeling reaction. Imaging was carried out on single-molecule level and each sample genome was de novo assembled from molecule data. Consensus genome maps were clustered into two alleles and diploid assemblies created. Genomes of patients were compared to parental genomes and the GRCh38 reference genome. SVs were inferred from de novo assemblies and genome comparisons with respect to quality scores, overall molecule coverage, fraction of molecules displaying the SV event, and chimeric DNA fragment mapping. Specific SV calls were compared to a set of > 160 human control samples (provided by Bionano Genomics) to filter against common SVs and potential artifacts. Filtered SVs were annotated using structural variant and gene databases. Employing WGOM we analyzed DNA molecules 300.000 bp long on average and achieved genomic coverage ranging from 90-132x corresponding to 330-480 Gbp. For instance, for one patient, we obtained 1751 insertions, 624 deletions, 77 inversions, 21 duplications, 1 intra- and 2 inter-chromosomal translocations before filtering. The majority of these events (78%) were inherited from both parents. 20% were inherited from either father or mother and 2% were generated de novo. As the family history of this patient was inconspicuous for tumor diseases, we removed all inherited events and filtered against common variants. This resulted in only two candidate de novo lesions: a heterozygous 129,495 bp deletion framed by inversions (chr9: 66,156,733-66,622,623) in a gene-less region and a heterozygous inverted 352,667 bp duplication (chr22: 15,522,454-15.875,120) that spanned the genes OR11H, POTEH, POTEH-AS1, LINC01297, DUXAP8, and BMS1P22. Of these genes DUXAP8 is an oncogenic non-coding RNA of the homeobox gene family that has been associated with increased tumor growth and poorer prognosis in a wide variety of somatic cancers. It functions as a regulator of transcription by binding to key components of the developmental regulator epigenetic polycomb repressive complex 2 and may thus account for additional presentations of the child (dwarfism, accelerated skeletal age, linguistic developmental delay, morphological traits). Our results indicate that WGOM is a useful technology to identify candidate SVs in children predisposed to cancer and developmental syndromes. Several candidates are currently being tested and the results will be presented. Disclosures No relevant conflicts of interest to declare.

Genome-Wide Analysis of Allele-Specific Expression Patterns in Seventeen Tissues of Korean Cattle (Hanwoo)

The functional hemizygosity could be caused by the MAE of a given gene and it can be one of the sources to affect the phenotypic variation in cattle. We aimed to identify MAE genes across the transcriptome in Korean cattle (Hanwoo). For three Hanwoo family trios, the transcriptome data of 17 tissues were generated in three offspring. Sixty-two MAE genes had a monoallelic expression in at least one tissue. Comparing genotypes among each family trio, the preferred alleles of 18 genes were identified (maternal expression, n = 9; paternal expression, n = 9). The MAE genes are involved in gene regulation, metabolic processes, and immune responses, and in particular, six genes encode transcription factors (FOXD2, FOXM1, HTATSF1, SCRT1, NKX6-2, and UBN1) with tissue-specific expression. In this study, we report genome-wide MAE genes in seventeen tissues of adult cattle. These results could help to elucidate epigenetic effects on phenotypic variation in Hanwoo.

Congenital X-Linked Myelodysplasia with Tetraploidy Is Associated with De Novo Germline C-Terminal Mutation of SEPT6, a Septin Filament Protein

Private germline mutations affecting hematopoiesis can cause progressive myelodysplasia and thus constitute pre-leukemic states. These can remain undetected, progressively transform and reveal themselves as infant leukemias, which can be linked to translocations involving the mixed lineage-leukemia (MLL) gene. Septin proteins play key roles in mammalian cell division and cytokinesis and are found as fusion partners of MLL in infant and early childhood acute myeloid leukemia (AML). We identified and describe a human germline disorder of septins in a newborn with myelodysplasia who required early hematopoietic stem cell transplant (HSCT) to prevent progressive disease. Materials & Methods: A Caucasian newborn male with no birth defects/malformations or suspicious family history presented with severe progressive neutropenia and was found to have bone marrow (BM) dysplasia with tetraploidy of myeloid progenitors. The patient developed unfavorable clonal aberrations (trisomy 7,8,9) and increased tetraploidy. Due to his progressive cytopenias and concern about risk of leukemic transformation, he underwent an allogeneic busulfan-cyclophosphamide/ATG conditioned DQ-mismatched unrelated HSCT at age 1 yo. He is currently 8 years post-HSCT with normal trilineage hematopoiesis (full donor chimerism), no graft versus host disease or any other non-hematological phenotype. To investigate the genetic etiology of this unique phenotype, we performed family trio germline exome/whole genome next-generation sequencing (NGS), and somatic pre-HSCT BM NGS for the index case. An established algorithm filtered for significant candidates following a de novo germline model. Immunohistochemical (IHC) staining of the pre- and post-HSCT BM biopsies for the candidate protein was performed. To validate the germline origin of the candidate mutation, we generated patient and control skin fibroblasts and induced pluripotent stem cells (iPSCs) that underwent fidelity testing by murine injection teratoma assays and 16-marker immunofluorescence (IF) staining. We then studied hematopoietic progenitor cells (HPCs) derived from iPSC-embryoid bodies (EB) in methylcellulose assays. To further determine the pathogenic nature of the mutation, we generated CRISPR/Cas9 knock-out of the human erythroleukemic cell line (TF-1) and studies these cells by cytomorphology, DNA and cell cycle assays. In-silico protein analysis of the candidate mutation and its effects on septin filament formation was performed. Results: Family trio and disease-tissue NGS identified a novel, germline C-terminal mutation in SEPT6, which was acquired de novo in the patient, and was not found in any database of common polymorphisms. IHC of pre-HSCT patient BM showed reduced Septin-6 staining in megakaryocyte and granulocyte precursors compared to post-HSCT and controls. Patient-derived iPSCs carried the mutation, were cytogenetically normal and bona-fide pluripotent by IF and teratoma assays. EB-derived HPCs from these cells recapitulated the patient's phenotype as they differentially failed to produce granulocyte vs erythroid colonies (fold-reductions CFU-M:8, CFU-G:36, CFU-GM:46, BFU-E:6, see Figure). Despite multiple approaches, SEPT6 CRISPR/Cas9 knock-out/in of the patient's mutation was not tolerated in iPSCs and human myeloid (granulo-/myelocytic) cell lines (HL-60, Molm-13, K562), and only tolerated in erythroid TF-1 cells. Analysis of SEPT6 knock-out TF-1 single-clone lines revealed a propensity to poly-nuclearity/lobation, as observed in the patient's BM. SEPT6 knock-in of C-terminal mutations caused cell death, consistent with existing literature. In silico protein analysis (incl. previously published crystallographic data) suggests that the mutation a) most likely modulates the key role of the coiled-coil SEPT6 domain in septin filament stabilization/bundling/bending, and thus deleteriously impacts cytokinesis, and b) perturbs the equilibrium of splice variants, possibly conferring tissue specificity. Conclusions: Mutation of the C-terminus of human SEPT6 causes aberrant cytokinesis in HPCs leading to a severe congenital neutropenia with tetraploidy and progressive myelodysplasia and cytogenetic aberrations. This report implicates a human germline disorder of SEPT6, and further investigations are required to elucidate the role septins in normal and disordered myelopoiesis. Figure. Figure. Disclosures Williams: Bluebird Bio: Research Funding.