DNN-Boost: Somatic mutation identification of tumor-only whole-exome sequencing data using deep neural network and XGBoost

Detection of somatic mutation in whole-exome sequencing data can help elucidate the mechanism of tumor progression. Most computational approaches require exome sequencing for both tumor and normal samples. However, it is more common to sequence exomes for tumor samples only without the paired normal samples. To include these types of data for extensive studies on the process of tumorigenesis, it is necessary to develop an approach for identifying somatic mutations using tumor exome sequencing data only. In this study, we designed a machine learning approach using Deep Neural Network (DNN) and XGBoost to identify somatic mutations in tumor-only exome sequencing data and we integrated this into a pipeline called DNN-Boost. The XGBoost algorithm is used to extract the features from the results of variant callers and these features are then fed into the DNN model as input. The XGBoost algorithm resolves issues of missing values and overfitting. We evaluated our proposed model and compared its performance with other existing benchmark methods. We noted that the DNN-Boost classification model outperformed the benchmark method in classifying somatic mutations from paired tumor-normal exome data and tumor-only exome data.

Effect of Loss-of-Function Genetic Variants in PCSK9 on Glycemic Traits, Neurocognitive Impairment, and Hepatobiliary Function

<strong>OBJECTIVE:</strong> To evaluate the association between <i>PCSK9 </i>predicted loss-of-function variants (pLoF) and glycemic traits, hepatobiliary function and neurocognitive traits. <p><strong>RESEARCH DESIGN AND METHODS:</strong> We identified carriers of <i>PCSK9</i> pLoF in UK Biobank exome sequencing data. We assessed the aggregate effects of these variants on lipid/lipoprotein traits, which served as a positive control. Association of <i>PCSK9 </i>pLoF carrier status and glycemic traits, hepatobiliary function, neurocognitive traits was then evaluated as a measure for adverse effects. </p> <p><strong>RESULTS:</strong> We identified 374 individuals with 41 pLoF variants. As expected, we found that <i>PCSK9</i> pLoF carriers had significantly lower LDL-C (<i>P</i> = 7.4 × 10^-55) and apoB levels (<i>P</i> = 7.6 × 10^-50) compared with noncarriers. However, we found no significant associations between pLoF carrier-status and glycemic traits, hepatobiliary function and neurocognitive traits (<i>P</i> > 0.05).</p> <p><strong>CONCLUSIONS</strong>: Our results do not support adverse effects of <i>PCSK9 </i>pLoF on glycemic traits, hepatobiliary function or neurocognitive traits.</p>

Effect of Loss-of-Function Genetic Variants in PCSK9 on Glycemic Traits, Neurocognitive Impairment, and Hepatobiliary Function

<strong>OBJECTIVE:</strong> To evaluate the association between <i>PCSK9 </i>predicted loss-of-function variants (pLoF) and glycemic traits, hepatobiliary function and neurocognitive traits. <p><strong>RESEARCH DESIGN AND METHODS:</strong> We identified carriers of <i>PCSK9</i> pLoF in UK Biobank exome sequencing data. We assessed the aggregate effects of these variants on lipid/lipoprotein traits, which served as a positive control. Association of <i>PCSK9 </i>pLoF carrier status and glycemic traits, hepatobiliary function, neurocognitive traits was then evaluated as a measure for adverse effects. </p> <p><strong>RESULTS:</strong> We identified 374 individuals with 41 pLoF variants. As expected, we found that <i>PCSK9</i> pLoF carriers had significantly lower LDL-C (<i>P</i> = 7.4 × 10^-55) and apoB levels (<i>P</i> = 7.6 × 10^-50) compared with noncarriers. However, we found no significant associations between pLoF carrier-status and glycemic traits, hepatobiliary function and neurocognitive traits (<i>P</i> > 0.05).</p> <p><strong>CONCLUSIONS</strong>: Our results do not support adverse effects of <i>PCSK9 </i>pLoF on glycemic traits, hepatobiliary function or neurocognitive traits.</p>

Secondary findings from whole-exome sequencing data in families with familial combined hyperlipidemia (FCHL)

Background During the interpretation of genome sequencing data, some types of secondary findings are identified that are located in genes that do not appear to be related to the causes of the primary disease. Although these are not the primary targets for evaluation, they have a high risk for some diseases different from the primary disease. Therefore, they can be vital for preventing and intervention from such disease. Results Here, we analyzed secondary findings obtained from WES in 6 families with FCHL disease who had an autosomal-dominant pattern based on their pedigrees. These finding are found in CDKAL1, ITGA2, FAM111A, WNK4, PTGIS, SCN10, TBX20, DCHS1, ANK2 and ABCA1 genes. Conclusions Secondary findings are very important and must be considered different variants from sequencing results in a diagnostic setting. Although we have considered these variants as secondary findings, some of them may be related to the primary disease.

Dnmt3a-mutated clonal hematopoiesis promotes osteoporosis

Osteoporosis is caused by an imbalance of osteoclasts and osteoblasts, occurring in close proximity to hematopoietic cells in the bone marrow. Recurrent somatic mutations that lead to an expanded population of mutant blood cells is termed clonal hematopoiesis of indeterminate potential (CHIP). Analyzing exome sequencing data from the UK Biobank, we found CHIP to be associated with increased incident osteoporosis diagnoses and decreased bone mineral density. In murine models, hematopoietic-specific mutations in Dnmt3a, the most commonly mutated gene in CHIP, decreased bone mass via increased osteoclastogenesis. Dnmt3a−/− demethylation opened chromatin and altered activity of inflammatory transcription factors. Bone loss was driven by proinflammatory cytokines, including Irf3-NF-κB–mediated IL-20 expression from Dnmt3a mutant macrophages. Increased osteoclastogenesis due to the Dnmt3a mutations was ameliorated by alendronate or IL-20 neutralization. These results demonstrate a novel source of osteoporosis-inducing inflammation.

Machine-learning algorithms predict breast cancer patient survival from UK Biobank whole-exome sequencing data

Aim: We tested whether machine-learning algorithm could find biomarkers predicting overall survival in breast cancer patients using blood-based whole-exome sequencing data. Materials & methods: Whole-exome sequencing data derived from 1181 female breast cancer patients within the UK Biobank was collected. We found feature genes (n = 50) regarding total mutation burden using the long short-term memory model. Then, we developed the XGBoost survival model with selected feature genes. Results: The XGBoost survival model performed acceptably, with a concordance index of 0.75 and a scaled Brier score of 0.146 in terms of overall survival prediction. The high-mutation group exhibited inferior overall survival compared with the low-mutation group in patients ≥56 years (log-rank test, p = 0.042). Conclusion: We showed that machine-learning algorithms can be used to predict overall survival in breast cancer patients from blood-based whole-exome sequencing data.