Multi-omic analysis reveals VEGFR2, PI3K, and JNK mediate the small molecule induction of human iPSC-derived cardiomyocyte proliferation.

Mammalian hearts lose their regenerative potential shortly after birth. Stimulating the proliferation of preexisting cardiomyocytes is a potential therapeutic strategy for cardiac damage. In a previous study, we identified 30 compounds that induced the bona-fide proliferation of human iPSC-derived cardiomyocytes (hiPSC-CM). Here, we selected five active compounds with diverse targets, including ALK5 and CB1R, and performed multi-omic analyses to identify common mechanisms mediating the cell cycle progression of hiPSC-CM. Transcriptome profiling revealed the top enriched pathways for all compounds including cell cycle, DNA repair, and kinesin pathways. Functional proteomic arrays found that the compounds collectively activated multiple receptor tyrosine kinases including ErbB2, IGF1R, and VEGFR2. Network analysis integrating common transcriptomic and proteomic signatures predicted that MAPK/PI3K pathways mediated compound responses. Furthermore, VEGFR2 negatively regulated endoreplication, enabling the completion of cell division. Thus, in this study, we applied high-content imaging and molecular profiling to establish mechanisms linking pro-proliferative agents to mechanisms of cardiomyocyte cell cycling.

DeepGSEA: explainable deep gene set enrichment analysis for single-cell transcriptomic data.

MOTIVATION: Gene set enrichment (GSE) analysis allows for an interpretation of gene expression through pre-defined gene set databases and is a critical step in understanding different phenotypes. With the rapid development of single-cell RNA sequencing (scRNA-seq) technology, GSE analysis can be performed on fine-grained gene expression data to gain a nuanced understanding of phenotypes of interest. However, with the cellular heterogeneity in single-cell gene profiles, current statistical GSE analysis methods sometimes fail to identify enriched gene sets. Meanwhile, deep learning has gained traction in applications like clustering and trajectory inference in single-cell studies due to its prowess in capturing complex data patterns. However, its use in GSE analysis remains limited, due to interpretability challenges. RESULTS: In this paper, we present DeepGSEA, an explainable deep gene set enrichment analysis approach which leverages the expressiveness of interpretable, prototype-based neural networks to provide an in-depth analysis of GSE. DeepGSEA learns the ability to capture GSE information through our designed classification tasks, and significance tests can be performed on each gene set, enabling the identification of enriched sets. The underlying distribution of a gene set learned by DeepGSEA can be explicitly visualized using the encoded cell and cellular prototype embeddings. We demonstrate the performance of DeepGSEA over commonly used GSE analysis methods by examining their sensitivity and specificity with four simulation studies. In addition, we test our model on three real scRNA-seq datasets and illustrate the interpretability of DeepGSEA by showing how its results can be explained. AVAILABILITY AND IMPLEMENTATION: https://github.com/Teddy-XiongGZ/DeepGSEA.

Novel Mouse Model of Myocardial Infarction, Plaque Rupture, and Stroke Shows Improved Survival With Myeloperoxidase Inhibition.

BACKGROUND: Thromboembolic events, including myocardial infarction (MI) or stroke, caused by the rupture or erosion of unstable atherosclerotic plaques are the leading cause of death worldwide. Although most mouse models of atherosclerosis develop lesions in the aorta and carotid arteries, they do not develop advanced coronary artery lesions. Moreover, they do not undergo spontaneous plaque rupture with MI and stroke or do so at such a low frequency that they are not viable experimental models to study late-stage thrombotic events or to identify novel therapeutic approaches for treating atherosclerotic disease. This has stymied the development of more effective therapeutic approaches for reducing these events beyond what has been achieved with aggressive lipid lowering. Here, we describe a diet-inducible mouse model that develops widespread advanced atherosclerosis in coronary, brachiocephalic, and carotid arteries with plaque rupture, MI, and stroke. METHODS: We characterized a novel mouse model with a C-terminal mutation in the scavenger receptor class B, type 1 (SR-BI), combined with Ldlr knockout (designated SR-BI∆CT/∆CT/Ldlr-/-). Mice were fed Western diet (WD) for 26 weeks and analyzed for MI and stroke. Coronary, brachiocephalic, and carotid arteries were analyzed for atherosclerotic lesions and indices of plaque stability. To validate the utility of this model, SR-BI∆CT/∆CT/Ldlr-/- mice were treated with the drug candidate AZM198, which inhibits myeloperoxidase, an enzyme produced by activated neutrophils that predicts rupture of human atherosclerotic lesions. RESULTS: SR-BI∆CT/∆CT/Ldlr-/- mice show high (>80%) mortality rates after 26 weeks of WD feeding because of major adverse cardiovascular events, including spontaneous plaque rupture with MI and stroke. Moreover, WD-fed SR-BI∆CT/∆CT/Ldlr-/- mice displayed elevated circulating high-sensitivity cardiac troponin I and increased neutrophil extracellular trap formation within lesions compared with control mice. Treatment of WD-fed SR-BI∆CT/∆CT/Ldlr-/- mice with AZM198 showed remarkable benefits, including >90% improvement in survival and >60% decrease in the incidence of plaque rupture, MI, and stroke, in conjunction with decreased circulating high-sensitivity cardiac troponin I and reduced neutrophil extracellular trap formation within lesions. CONCLUSIONS: WD-fed SR-BI∆CT/∆CT/Ldlr-/- mice more closely replicate late-stage clinical events of advanced human atherosclerotic disease than previous models and can be used to identify and test potential new therapeutic agents to prevent major adverse cardiac events.

Loss of TET2 increases B-1 cell number and IgM production while limiting CDR3 diversity.

Recent studies have demonstrated a role for Ten-Eleven Translocation-2 (TET2), an epigenetic modulator, in regulating germinal center formation and plasma cell differentiation in B-2 cells, yet the role of TET2 in regulating B-1 cells is largely unknown. Here, B-1 cell subset numbers, IgM production, and gene expression were analyzed in mice with global knockout of TET2 compared to wildtype (WT) controls. Results revealed that TET2-KO mice had elevated numbers of B-1a and B-1b cells in their primary niche, the peritoneal cavity, as well as in the bone marrow (B-1a) and spleen (B-1b). Consistent with this finding, circulating IgM, but not IgG, was elevated in TET2-KO mice compared to WT. Analysis of bulk RNASeq of sort purified peritoneal B-1a and B-1b cells revealed reduced expression of heavy and light chain immunoglobulin genes, predominantly in B-1a cells from TET2-KO mice compared to WT controls. As expected, the expression of IgM transcripts was the most abundant isotype in B-1 cells. Yet, only in B-1a cells there was a significant increase in the proportion of IgM transcripts in TET2-KO mice compared to WT. Analysis of the CDR3 of the BCR revealed an increased abundance of replicated CDR3 sequences in B-1 cells from TET2-KO mice, which was more clearly pronounced in B-1a compared to B-1b cells. V-D-J usage and circos plot analysis of V-J combinations showed enhanced usage of VH11 and VH12 pairings. Taken together, our study is the first to demonstrate that global loss of TET2 increases B-1 cell number and IgM production and reduces CDR3 diversity, which could impact many biological processes and disease states that are regulated by IgM.

Genome-scale chromatin binding dynamics of RNA Polymerase II general transcription machinery components.

A great deal of work has revealed, in structural detail, the components of the preinitiation complex (PIC) machinery required for initiation of mRNA gene transcription by RNA polymerase II (Pol II). However, less-well understood are the in vivo PIC assembly pathways and their kinetics, an understanding of which is vital for determining how rates of in vivo RNA synthesis are established. We used competition ChIP in budding yeast to obtain genome-scale estimates of the residence times for five general transcription factors (GTFs): TBP, TFIIA, TFIIB, TFIIE and TFIIF. While many GTF-chromatin interactions were short-lived ( < 1 min), there were numerous interactions with residence times in the range of several minutes. Sets of genes with a shared function also shared similar patterns of GTF kinetic behavior. TFIIE, a GTF that enters the PIC late in the assembly process, had residence times correlated with RNA synthesis rates. The datasets and results reported here provide kinetic information for most of the Pol II-driven genes in this organism, offering a rich resource for exploring the mechanistic relationships between PIC assembly, gene regulation, and transcription.

DARDN: A Deep-Learning Approach for CTCF Binding Sequence Classification and Oncogenic Regulatory Feature Discovery.

Characterization of gene regulatory mechanisms in cancer is a key task in cancer genomics. CCCTC-binding factor (CTCF), a DNA binding protein, exhibits specific binding patterns in the genome of cancer cells and has a non-canonical function to facilitate oncogenic transcription programs by cooperating with transcription factors bound at flanking distal regions. Identification of DNA sequence features from a broad genomic region that distinguish cancer-specific CTCF binding sites from regular CTCF binding sites can help find oncogenic transcription factors in a cancer type. However, the presence of long DNA sequences without localization information makes it difficult to perform conventional motif analysis. Here, we present DNAResDualNet (DARDN), a computational method that utilizes convolutional neural networks (CNNs) for predicting cancer-specific CTCF binding sites from long DNA sequences and employs DeepLIFT, a method for interpretability of deep learning models that explains the model's output in terms of the contributions of its input features. The method is used for identifying DNA sequence features associated with cancer-specific CTCF binding. Evaluation on DNA sequences associated with CTCF binding sites in T-cell acute lymphoblastic leukemia (T-ALL) and other cancer types demonstrates DARDN's ability in classifying DNA sequences surrounding cancer-specific CTCF binding from control constitutive CTCF binding and identifying sequence motifs for transcription factors potentially active in each specific cancer type. We identify potential oncogenic transcription factors in T-ALL, acute myeloid leukemia (AML), breast cancer (BRCA), colorectal cancer (CRC), lung adenocarcinoma (LUAD), and prostate cancer (PRAD). Our work demonstrates the power of advanced machine learning and feature discovery approach in finding biologically meaningful information from complex high-throughput sequencing data.

Genome-scale chromatin interaction dynamic measurements for key components of the RNA Pol II general transcription machinery.

Background: A great deal of work has revealed in structural detail the components of the machinery responsible for mRNA gene transcription initiation. These include the general transcription factors (GTFs), which assemble at promoters along with RNA Polymerase II (Pol II) to form a preinitiation complex (PIC) aided by the activities of cofactors and site-specific transcription factors (TFs). However, less well understood are the in vivo PIC assembly pathways and their kinetics, an understanding of which is vital for determining on a mechanistic level how rates of in vivo RNA synthesis are established and how cofactors and TFs impact them. Results: We used competition ChIP to obtain genome-scale estimates of the residence times for five GTFs: TBP, TFIIA, TFIIB, TFIIE and TFIIF in budding yeast. While many GTF-chromatin interactions were short-lived (< 1 min), there were numerous interactions with residence times in the several minutes range. Sets of genes with a shared function also shared similar patterns of GTF kinetic behavior. TFIIE, a GTF that enters the PIC late in the assembly process, had residence times correlated with RNA synthesis rates. Conclusions: The datasets and results reported here provide kinetic information for most of the Pol II-driven genes in this organism and therefore offer a rich resource for exploring the mechanistic relationships between PIC assembly, gene regulation, and transcription. The relationships between gene function and GTF dynamics suggest that shared sets of TFs tune PIC assembly kinetics to ensure appropriate levels of expression.

ProtoCell4P: an explainable prototype-based neural network for patient classification using single-cell RNA-seq.

MOTIVATION: The rapid advance in single-cell RNA sequencing (scRNA-seq) technology over the past decade has provided a rich resource of gene expression profiles of single cells measured on patients, facilitating the study of many biological questions at the single-cell level. One intriguing research is to study the single cells which play critical roles in the phenotypes of patients, which has the potential to identify those cells and genes driving the disease phenotypes. To this end, deep learning models are expected to well encode the single-cell information and achieve precise prediction of patients' phenotypes using scRNA-seq data. However, we are facing critical challenges in designing deep learning models for classifying patient samples due to (i) the samples collected in the same dataset contain a variable number of cells-some samples might only have hundreds of cells sequenced while others could have thousands of cells, and (ii) the number of samples available is typically small and the expression profile of each cell is noisy and extremely high-dimensional. Moreover, the black-box nature of existing deep learning models makes it difficult for the researchers to interpret the models and extract useful knowledge from them. RESULTS: We propose a prototype-based and cell-informed model for patient phenotype classification, termed ProtoCell4P, that can alleviate problems of the sample scarcity and the diverse number of cells by leveraging the cell knowledge with representatives of cells (called prototypes), and precisely classify the patients by adaptively incorporating information from different cells. Moreover, this classification process can be explicitly interpreted by identifying the key cells for decision making and by further summarizing the knowledge of cell types to unravel the biological nature of the classification. Our approach is explainable at the single-cell resolution which can identify the key cells in each patient's classification. The experimental results demonstrate that our proposed method can effectively deal with patient classifications using single-cell data and outperforms the existing approaches. Furthermore, our approach is able to uncover the association between cell types and biological classes of interest from a data-driven perspective. AVAILABILITY AND IMPLEMENTATION: https://github.com/Teddy-XiongGZ/ProtoCell4P.

Interpretable meta-learning of multi-omics data for survival analysis and pathway enrichment.

MOTIVATION: Despite the success of recent machine learning algorithms' applications to survival analysis, their black-box nature hinders interpretability, which is arguably the most important aspect. Similarly, multi-omics data integration for survival analysis is often constrained by the underlying relationships and correlations that are rarely well understood. The goal of this work is to alleviate the interpretability problem in machine learning approaches for survival analysis and also demonstrate how multi-omics data integration improves survival analysis and pathway enrichment. We use meta-learning, a machine-learning algorithm that is trained on a variety of related datasets and allows quick adaptations to new tasks, to perform survival analysis and pathway enrichment on pan-cancer datasets. In recent machine learning research, meta-learning has been effectively used for knowledge transfer among multiple related datasets. RESULTS: We use meta-learning with Cox hazard loss to show that the integration of TCGA pan-cancer data increases the performance of survival analysis. We also apply advanced model interpretability method called DeepLIFT (Deep Learning Important FeaTures) to show different sets of enriched pathways for multi-omics and transcriptomics data. Our results show that multi-omics cancer survival analysis enhances performance compared with using transcriptomics or clinical data alone. Additionally, we show a correlation between variable importance assignment from DeepLIFT and gene coenrichment, suggesting that genes with higher and similar contribution scores are more likely to be enriched together in the same enrichment sets. AVAILABILITY AND IMPLEMENTATION: https://github.com/berkuva/TCGA-omics-integration.

A Step Toward Personalized Surgical Decision Making: Machine Learning Predicts 1 Versus Numerous Melanoma Lymph Node Metastases Using RNA-sequencing.

OBJECTIVE: Develop a predictive model to identify patients with 1 pathologic lymph node (pLN) versus >1 pLN using machine learning applied to gene expression profiles and clinical data as input variables. BACKGROUND: Standard management for clinically detected melanoma lymph node metastases is complete therapeutic LN dissection (TLND). However, >40% of patients with a clinically detected melanoma lymph node will only have 1 pLN on final review. Recent data suggest that targeted excision of just the single enlarged LN may provide excellent regional control, with less morbidity than TLND. The selection of patients for less morbid surgery requires accurate identification of those with only 1 pLN. METHODS: The Cancer Genome Atlas database was used to identify patients who underwent TLND for melanoma. Pathology reports in The Cancer Genome Atlas were reviewed to identify the number of pLNs. Patients were included for machine learning analyses if RNA sequencing data were available from a pLN. After feature selection, the top 20 gene expression and clinical input features were used to train a ridge logistic regression model to predict patients with 1 pLN versus >1 pLN using 10-fold cross-validation on 80% of samples. The model was then tested on the remaining holdout samples. RESULTS: A total of 153 patients met inclusion criteria: 64 with one pLN (42%) and 89 with >1 pLNs (58%). Feature selection identified 1 clinical (extranodal extension) and 19 gene expression variables used to predict patients with 1 pLN versus >1 pLN. The ridge logistic regression model identified patient groups with an accuracy of 90% and an area under the receiver operating characteristic curve of 0.97. CONCLUSIONS: Gene expression profiles together with clinical variables can distinguish melanoma metastasis patients with 1 pLN versus >1 pLN. Future models trained using positron emission tomography/computed tomography imaging, gene expression, and relevant clinical variables may further improve accuracy and may predict patients who can be managed with a targeted LN excision rather than a complete TLND.

Generalization Performance of Quantum Metric Learning Classifiers.

Quantum computing holds great promise for a number of fields including biology and medicine. A major application in which quantum computers could yield advantage is machine learning, especially kernel-based approaches. A recent method termed quantum metric learning, in which a quantum embedding which maximally separates data into classes is learned, was able to perfectly separate ant and bee image training data. The separation is achieved with an intrinsically quantum objective function and the overall approach was shown to work naturally as a hybrid classical-quantum computation enabling embedding of high dimensional feature data into a small number of qubits. However, the ability of the trained classifier to predict test sample data was never assessed. We assessed the performance of quantum metric learning on test ants and bees image data as well as breast cancer clinical data. We applied the original approach as well as variants in which we performed principal component analysis (PCA) on the feature data to reduce its dimensionality for quantum embedding, thereby limiting the number of model parameters. If the degree of dimensionality reduction was limited and the number of model parameters was constrained to be far less than the number of training samples, we found that quantum metric learning was able to accurately classify test data.

Inhibition of DYRK1a Enhances Cardiomyocyte Cycling After Myocardial Infarction.

BACKGROUND: DYRK1a (dual-specificity tyrosine phosphorylation-regulated kinase 1a) contributes to the control of cycling cells, including cardiomyocytes. However, the effects of inhibition of DYRK1a on cardiac function and cycling cardiomyocytes after myocardial infarction (MI) remain unknown. METHODS: We investigated the impacts of pharmacological inhibition and conditional genetic ablation of DYRK1a on endogenous cardiomyocyte cycling and left ventricular systolic function in ischemia-reperfusion (I/R) MI using αMHC-MerDreMer-Ki67p-RoxedCre::Rox-Lox-tdTomato-eGFP (RLTG) (denoted αDKRC::RLTG) and αMHC-Cre::Fucci2aR::DYRK1aflox/flox mice. RESULTS: We observed that harmine, an inhibitor of DYRK1a, improved left ventricular ejection fraction (39.5±1.6% and 29.1±1.6%, harmine versus placebo, respectively), 2 weeks after I/R MI. Harmine also increased cardiomyocyte cycling after I/R MI in αDKRC::RLTG mice, 10.8±1.5 versus 24.3±2.6 enhanced Green Fluorescent Protein (eGFP)+ cardiomyocytes, placebo versus harmine, respectively, P=1.0×10-3. The effects of harmine on left ventricular ejection fraction were attenuated in αDKRC::DTA mice that expressed an inducible diphtheria toxin in adult cycling cardiomyocytes. The conditional cardiomyocyte-specific genetic ablation of DYRK1a in αMHC-Cre::Fucci2aR::DYRK1aflox/flox (denoted DYRK1a k/o) mice caused cardiomyocyte hyperplasia at baseline (210±28 versus 126±5 cardiomyocytes per 40× field, DYRK1a k/o versus controls, respectively, P=1.7×10-2) without changes in cardiac function compared with controls, or compensatory changes in the expression of other DYRK isoforms. After I/R MI, DYRK1a k/o mice had improved left ventricular function (left ventricular ejection fraction 41.8±2.2% and 26.4±0.8%, DYRK1a k/o versus control, respectively, P=3.7×10-2). RNAseq of cardiomyocytes isolated from αMHC-Cre::Fucci2aR::DYRK1aflox/flox and αMHC-Cre::Fucci2aR mice after I/R MI or Sham surgeries identified enrichment in mitotic cell cycle genes in αMHC-Cre::Fucci2aR::DYRK1aflox/flox compared with αMHC-Cre::Fucci2aR. CONCLUSIONS: The pharmacological inhibition or cardiomyocyte-specific ablation of DYRK1a caused baseline hyperplasia and improved cardiac function after I/R MI, with an increase in cell cycle gene expression, suggesting the inhibition of DYRK1a may serve as a therapeutic target to treat MI.

Associations of immune cell homing gene signatures and infiltrates of lymphocyte subsets in human melanomas: discordance with CD163+ myeloid cell infiltrates.

BACKGROUND: Immune cells in the tumor microenvironment have prognostic value. In preclinical models, recruitment and infiltration of these cells depends on immune cell homing (ICH) genes such as chemokines, cell adhesion molecules, and integrins. We hypothesized ICH ligands CXCL9-11 and CCL2-5 would be associated with intratumoral T-cells, while CXCL13 would be more associated with B-cell infiltrates. METHODS: Samples of human melanoma were submitted for gene expression analysis and immune cells identified by immunohistochemistry. Associations between the two were evaluated with unsupervised hierarchical clustering using correlation matrices from Spearman rank tests. Univariate analysis performed Mann-Whitney tests. RESULTS: For 119 melanoma specimens, analysis of 78 ICH genes revealed association among genes with nonspecific increase of multiple immune cell subsets: CD45+, CD8+ and CD4+ T-cells, CD20+ B-cells, CD138+ plasma cells, and CD56+ NK-cells. ICH genes most associated with these infiltrates included ITGB2, ITGAL, CCL19, CXCL13, plus receptor/ligand pairs CXCL9 and CXCL10 with CXCR3; CCL4 and CCL5 with CCR5. This top ICH gene expression signature was also associated with genes representing immune-activation and effector function. In contrast, CD163+ M2-macrophages was weakly associated with a different ICH gene signature. CONCLUSION: These data do not support our hypothesis that each immune cell subset is uniquely associated with specific ICH genes. Instead, a larger set of ICH genes identifies melanomas with concordant infiltration of B-cell and T-cell lineages, while CD163+ M2-macrophage infiltration suggesting alternate mechanisms for their recruitment. Future studies should explore the extent ICH gene signature contributes to tertiary lymphoid structures or cross-talk between homing pathways.

IGF1R and Src inhibition induce synergistic cytotoxicity in HNSCC through inhibition of FAK.

Head and neck cancer is the sixth most common cancer worldwide with a 5-year survival of only 65%. Targeting compensatory signaling pathways may improve therapeutic responses and combat resistance. Utilizing reverse phase protein arrays (RPPA) to assess the proteome and explore mechanisms of synergistic growth inhibition in HNSCC cell lines treated with IGF1R and Src inhibitors, BMS754807 and dasatinib, respectively, we identified focal adhesion signaling as a critical node. Focal Adhesion Kinase (FAK) and Paxillin phosphorylation were decreased as early as 15 min after treatment, and treatment with a FAK inhibitor, PF-562,271, was sufficient to decrease viability in vitro. Treatment of 3D spheroids demonstrated robust cytotoxicity suggesting that the combination of BMS754807 and dasatinib is effective in multiple experimental models. Furthermore, treatment with BMS754807 and dasatinib significantly decreased cell motility, migration, and invasion in multiple HNSCC cell lines. Most strikingly, treatment with BMS754807 and dasatinib, or a FAK inhibitor alone, significantly increased cleaved-PARP in human ex-vivo HNSCC patient tissues demonstrating a potential clinical utility for targeting FAK or the combined targeting of the IGF1R with Src. This ex-vivo result further confirms FAK as a vital signaling node of this combinatorial treatment and demonstrates therapeutic potential for targeting FAK in HNSCC patients.

Upregulation of human endogenous retrovirus-K (HML-2) mRNAs in hepatoblastoma: Identification of potential new immunotherapeutic targets and biomarkers.

PURPOSE: Hepatoblastoma is the most common liver malignancy in children. In order to advance therapy against hepatoblastoma, novel immunologic targets and biomarkers are needed. Our purpose in this investigation is to examine hepatoblastoma transcriptomes for the expression of a class of genomic elements known as Human Endogenous Retrovirus (HERVs). HERVs are abundant in the human genome and are biologically active elements that have been associated with multiple malignancies and proposed as immunologic targets in a subset of tumors. A sub-family of HERVs, HERV-K(HML-2) (HERV-K), have been shown to be tightly regulated in fetal development, making investigation of these elements in pediatric tumors paramount. METHODS: We first created a HERVK-FASTA file utilizing 91 previously described HML-2 proviruses. We then concatenated the file onto the GRCh38.95 cDNA library from Ensembl. We used this reference database to evaluate existing RNA-seq data from 10 hepatoblastoma tumors and 3 normal liver controls (GEO accession ID: GSE8977575). Quantification and differential proviral expression analysis between hepatoblastoma and normal liver controls was performed using the pseudo-alignment program Salmon and DESeq2, respectively. RESULTS: HERV-K mRNA was expressed in hepatoblastoma from multiple proviral loci. All expressed HERV-K proviral loci were upregulated in hepatoblastoma compared to normal liver controls. Five HERV-K proviruses (1q21.3, 3q27.2, 7q22.2, 12q24.33 and 17p13.1) were significantly differentially expressed (p-adjusted value <0.05, |log2 fold change| > 1.5) across conditions. The provirus at 17p13.1 had an approximately 300-fold increased expression in hepatoblastoma as compared to normal liver. This was in part due to the near absence of HERV-K mRNA at the 17p13.1 locus in fully differentiated liver samples. CONCLUSIONS: Our investigation demonstrates that HERV-K is expressed from multiple loci in hepatoblastoma and that the expression is increased for several proviruses compared to normal liver controls. Our results suggest that HERV-K mRNA expression may be useful as a biomarker in hepatoblastoma, given the large differential expression profiles in hepatoblastoma, with very low mRNA levels in liver control samples.

A Machine Learning Model Utilizing a Novel SNP Shows Enhanced Prediction of Coronary Artery Disease Severity.

Background: Machine learning (ML) has emerged as a powerful approach for predicting outcomes based on patterns and inferences. Improving prediction of severe coronary artery disease (CAD) has the potential for personalizing prevention and treatment strategies and for identifying individuals that may benefit from cardiac catheterization. We developed a novel ML approach combining traditional cardiac risk factors (CRF) with a single nucleotide polymorphism (SNP) in a gene associated with human CAD (ID3 rs11574) to enhance prediction of CAD severity; Methods: ML models incorporating CRF along with ID3 genotype at rs11574 were evaluated. The most predictive model, a deep neural network, was used to classify patients into high (>32) and low level (≤32) Gensini severity score. This model was trained on 325 and validated on 82 patients. Prediction performance of the model was summarized by a confusion matrix and area under the receiver operating characteristics curve (ROC-AUC); and Results: Our neural network predicted severity score with 81% and 87% accuracy for the low and the high groups respectively with an ROC-AUC of 0.84 for 82 patients in the test group. The addition of ID3 rs11574 to CRF significantly enhanced prediction accuracy from 65% to 81% in the low group, and 72% to 84% in the high group. Age, high-density lipoprotein (HDL), and systolic blood pressure were the top 3 contributors in predicting severity score; Conclusions: Our neural network including ID3 rs11574 improved prediction of CAD severity over use of Framingham score, which may potentially be helpful for clinical decision making in patients at increased risk of complications from coronary angiography.

COCOA: coordinate covariation analysis of epigenetic heterogeneity.

A key challenge in epigenetics is to determine the biological significance of epigenetic variation among individuals. We present Coordinate Covariation Analysis (COCOA), a computational framework that uses covariation of epigenetic signals across individuals and a database of region sets to annotate epigenetic heterogeneity. COCOA is the first such tool for DNA methylation data and can also analyze any epigenetic signal with genomic coordinates. We demonstrate COCOA's utility by analyzing DNA methylation, ATAC-seq, and multi-omic data in supervised and unsupervised analyses, showing that COCOA provides new understanding of inter-sample epigenetic variation. COCOA is available on Bioconductor ( http://bioconductor.org/packages/COCOA ).

Implementation of a Hamming distance-like genomic quantum classifier using inner products on ibmqx2 and ibmq_16_melbourne.

Motivated by the problem of classifying individuals with a disease versus controls using a functional genomic attribute as input, we present relatively efficient general purpose inner product-based kernel classifiers to classify the test as a normal or disease sample. We encode each training sample as a string of 1 s (presence) and 0 s (absence) representing the attribute's existence across ordered physical blocks of the subdivided genome. Having binary-valued features allows for highly efficient data encoding in the computational basis for classifiers relying on binary operations. Given that a natural distance between binary strings is Hamming distance, which shares properties with bit-string inner products, our two classifiers apply different inner product measures for classification. The active inner product (AIP) is a direct dot product-based classifier whereas the symmetric inner product (SIP) classifies upon scoring correspondingly matching genomic attributes. SIP is a strongly Hamming distance-based classifier generally applicable to binary attribute-matching problems whereas AIP has general applications as a simple dot product-based classifier. The classifiers implement an inner product between N = 2 n dimension test and train vectors using n Fredkin gates while the training sets are respectively entangled with the class-label qubit, without use of an ancilla. Moreover, each training class can be composed of an arbitrary number m of samples that can be classically summed into one input string to effectively execute all test-train inner products simultaneously. Thus, our circuits require the same number of qubits for any number of training samples and are O ( log N ) in gate complexity after the states are prepared. Our classifiers were implemented on ibmqx2 (IBM-Q-team 2019b) and ibmq_16_melbourne (IBM-Q-team 2019a). The latter allowed encoding of 64 training features across the genome.