MOSCATO: a supervised approach for analyzing multi-Omic single-Cell data.

BACKGROUND: Advancements in genomic sequencing continually improve personalized medicine, and recent breakthroughs generate multimodal data on a cellular level. We introduce MOSCATO, a technique for selecting features across multimodal single-cell datasets that relate to clinical outcomes. We summarize the single-cell data using tensors and perform regularized tensor regression to return clinically-associated variable sets for each 'omic' type. RESULTS: Robustness was assessed over simulations based on available single-cell simulation methods, and applicability was assessed through an example using CITE-seq data to detect genes associated with leukemia. We find that MOSCATO performs favorably in selecting network features while also shown to be applicable to real multimodal single-cell data. CONCLUSIONS: MOSCATO is a useful analytical technique for supervised feature selection in multimodal single-cell data. The flexibility of our approach enables future extensions on distributional assumptions and covariate adjustments.

Influence of Tyrosine Kinase Inhibition on Organic Anion Transporting Polypeptide 1B3-Mediated Uptake.

The organic anion transporting polypeptide family member (OATP) 1B3 is a hepatic uptake transporter that has a broad substrate recognition and plays a significant role in regulating elimination of endogenous biomolecules or xenobiotics. OATP1B3 works in tandem with OATP1B1, with which it shares approximately 80% sequence homology and a high degree of substrate overlap. Despite some substrates being recognized solely by OATP1B3, its ability to compensate for loss of OATP1B1-mediated elimination and recognition by regulatory agencies, little is known about OATP1B3 regulatory factors and how they are involved with drug-drug interaction. It was recently discovered that OATP1B1 function is mediated by the activity of a particular tyrosine kinase that is sensitive to a variety of tyrosine kinase inhibitors (TKIs). This study reports that OATP1B3 is similarly regulated, as at least 50% of its activity is reduced by 20 US Food and Drug Administration -approved TKIs. Nilotinib was assessed as the most potent OATP1B3 inhibitor among the investigated TKIs, which can occur at clinically relevant concentrations and acted predominantly through noncompetitive inhibition without impacting membrane expression. Finally, OATP1B3 function was determined to be sensitive to the knockdown of the Lck/Yes novel tyrosine kinase that is sensitive to nilotinib and has been previously implicated in mediating OATP1B1 activity. Collectively, our findings identify tyrosine kinase activity as a major regulator of OATP1B3 function which is sensitive to kinase inhibition. Given that OATP1B1 is similarly regulated, simultaneous disruption of these transporters can have drastic effects on systemic drug concentrations, which would promote adverse events. SIGNIFICANCE STATEMENT: The organic anion transporting polypeptide family member (OATP) 1B3 is a facilitator of hepatic drug elimination, although much is unknown of how OATP1B3 activity is mediated, or how such regulators contribute to drug-drug interactions. This study reports that OATP1B3 activity is dependent on the Lck/Yes novel tyrosine kinase, which is sensitive to numerous tyrosine kinase inhibitors. These findings provide insight into the occurrence of many clinical drug-drug interactions, and a rationale for future study of tyrosine kinases regulating drug disposition.

SuMO-Fil: Supervised multi-omic filtering prior to performing network analysis.

Multi-omic analyses that integrate many high-dimensional datasets often present significant deficiencies in statistical power and require time consuming computations to execute the analytical methods. We present SuMO-Fil to remedy against these issues which is a pre-processing method for Supervised Multi-Omic Filtering that removes variables or features considered to be irrelevant noise. SuMO-Fil is intended to be performed prior to downstream analyses that detect supervised gene networks in sparse settings. We accomplish this by implementing variable filters based on low similarity across the datasets in conjunction with low similarity with the outcome. This approach can improve accuracy, as well as reduce run times for a variety of computationally expensive downstream analyses. This method has applications in a setting where the downstream analysis may include sparse canonical correlation analysis. Filtering methods specifically for cluster and network analysis are introduced and compared by simulating modular networks with known statistical properties. The SuMO-Fil method performs favorably by eliminating non-network features while maintaining important biological signal under a variety of different signal settings as compared to popular filtering techniques based on low means or low variances. We show that the speed and accuracy of methods such as supervised sparse canonical correlation are increased after using SuMO-Fil, thus greatly improving the scalability of these approaches.

Gene-set integrative analysis of multi-omics data using tensor-based association test.

MOTIVATION: Facilitated by technological advances and the decrease in costs, it is feasible to gather subject data from several omics platforms. Each platform assesses different molecular events, and the challenge lies in efficiently analyzing these data to discover novel disease genes or mechanisms. A common strategy is to regress the outcomes on all omics variables in a gene set. However, this approach suffers from problems associated with high-dimensional inference. RESULTS: We introduce a tensor-based framework for variable-wise inference in multi-omics analysis. By accounting for the matrix structure of an individual's multi-omics data, the proposed tensor methods incorporate the relationship among omics effects, reduce the number of parameters, and boost the modeling efficiency. We derive the variable-specific tensor test and enhance computational efficiency of tensor modeling. Using simulations and data applications on the Cancer Cell Line Encyclopedia (CCLE), we demonstrate our method performs favorably over baseline methods and will be useful for gaining biological insights in multi-omics analysis. AVAILABILITY AND IMPLEMENTATION: R function and instruction are available from the authors' website: https://www4.stat.ncsu.edu/~jytzeng/Software/TR.omics/TRinstruction.pdf. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Automated Computational Detection of Interstitial Fibrosis, Tubular Atrophy, and Glomerulosclerosis.

BACKGROUND: Interstitial fibrosis, tubular atrophy (IFTA), and glomerulosclerosis are indicators of irrecoverable kidney injury. Modern machine learning (ML) tools have enabled robust, automated identification of image structures that can be comparable with analysis by human experts. ML algorithms were developed and tested for the ability to replicate the detection and quantification of IFTA and glomerulosclerosis that renal pathologists perform. METHODS: A renal pathologist annotated renal biopsy specimens from 116 whole-slide images (WSIs) for IFTA and glomerulosclerosis. A total of 79 WSIs were used for training different configurations of a convolutional neural network (CNN), and 17 and 20 WSIs were used as internal and external testing cases, respectively. The best model was compared against the input of four renal pathologists on 20 new testing slides. Further, for 87 testing biopsy specimens, IFTA and glomerulosclerosis measurements made by pathologists and the CNN were correlated to patient outcome using classic statistical tools. RESULTS: The best average performance across all image classes came from a DeepLab version 2 network trained at 40× magnification. IFTA and glomerulosclerosis percentages derived from this CNN achieved high levels of agreement with four renal pathologists. The pathologist- and CNN-based analyses of IFTA and glomerulosclerosis showed statistically significant and equivalent correlation with all patient-outcome variables. CONCLUSIONS: ML algorithms can be trained to replicate the IFTA and glomerulosclerosis assessment performed by renal pathologists. This suggests computational methods may be able to provide a standardized approach to evaluate the extent of chronic kidney injury in situations in which renal-pathologist time is restricted or unavailable.

Balanced Functional Module Detection in genomic data.

Motivation: High-dimensional genomic data can be analyzed to understand the effects of variables on a target variable such as a clinical outcome. For understanding the underlying biological mechanism affecting the target, it is important to discover the complete set of relevant variables. Thus variable selection is a primary goal, which differs from a prediction criterion. Of special interest are functional modules, cooperating sets of variables affecting the target which can be characterized by a graph. In applications such as social networks, the concept of balance in undirected signed graphs characterizes the consistency of associations within the network. This property requires that the module variables have a joint effect on the target outcome with no internal conflict, an efficiency that may be applied to biological networks. Results: In this paper, we model genomic variables in signed undirected graphs for applications where the set of predictor variables influences an outcome. Consequences of the balance property are exploited to implement a new module discovery algorithm, balanced Functional Module Detection (bFMD), which selects a subset of variables from high-dimensional data that compose a balanced functional module. Our bFMD algorithm performed favorably in simulations as compared to other module detection methods. Additionally, bFMD detected interpretable results in an application using RNA-seq data obtained from subjects with Uterine Corpus Endometrial Carcinoma using the percentage of tumor invasion as the outcome of interest. The variables selected by bFMD have improved interpretability due to the logical consistency afforded by the balance property. Supplementary information: Supplementary data are available at Bioinformatics Advances online.

Identification of supervised and sparse functional genomic pathways.

Functional pathways involve a series of biological alterations that may result in the occurrence of many diseases including cancer. With the availability of various "omics" technologies it becomes feasible to integrate information from a hierarchy of biological layers to provide a more comprehensive understanding to the disease. In many diseases, it is believed that only a small number of networks, each relatively small in size, drive the disease. Our goal in this study is to develop methods to discover these functional networks across biological layers correlated with the phenotype. We derive a novel Network Summary Matrix (NSM) that highlights potential pathways conforming to least squares regression relationships. An algorithm called Decomposition of Network Summary Matrix via Instability (DNSMI) involving decomposition of NSM using instability regularization is proposed. Simulations and real data analysis from The Cancer Genome Atlas (TCGA) program will be shown to demonstrate the performance of the algorithm.

A novel characterization of the generalized family wise error rate using empirical null distributions.

We present a novel characterization of the generalized family wise error rate: kFWER. The interpretation allows researchers to view kFWER as a function of the test statistics rather than current methods based on p-values. Using this interpretation we present several theorems and methods (parametric and non-parametric) for estimating kFWER in various data settings. With this version of kFWER, researchers will have an estimate of kFWER in addition to knowing what tests are significant at the estimated kFWER. Additionally, we present methods that use empirical null distributions in place of parametric distributions in standard p-value kFWER controlling schemes. These advancements represent an improvement over common kFWER methods which are based on parametric assumptions and merely report the tests that are significant under a given value for kFWER.

A purine nucleotide biosynthesis enzyme guanosine monophosphate reductase is a suppressor of melanoma invasion.

Melanoma is one of the most aggressive types of human cancers, and the mechanisms underlying melanoma invasive phenotype are not completely understood. Here, we report that expression of guanosine monophosphate reductase (GMPR), an enzyme involved in de novo biosynthesis of purine nucleotides, was downregulated in the invasive stages of human melanoma. Loss- and gain-of-function experiments revealed that GMPR downregulates the amounts of several GTP-bound (active) Rho-GTPases and suppresses the ability of melanoma cells to form invadopodia, degrade extracellular matrix, invade in vitro, and grow as tumor xenografts in vivo. Mechanistically, we demonstrated that GMPR partially depletes intracellular GTP pools. Pharmacological inhibition of de novo GTP biosynthesis suppressed whereas addition of exogenous guanosine increased invasion of melanoma cells as well as cells from other cancer types. Our data identify GMPR as a melanoma invasion suppressor and establish a link between guanosine metabolism and Rho-GTPase-dependent melanoma cell invasion.

Facilitates chromatin transcription complex is an "accelerator" of tumor transformation and potential marker and target of aggressive cancers.

The facilitates chromatin transcription (FACT) complex is involved in chromatin remodeling during transcription, replication, and DNA repair. FACT was previously considered to be ubiquitously expressed and not associated with any disease. However, we discovered that FACT is the target of a class of anticancer compounds and is not expressed in normal cells of adult mammalian tissues, except for undifferentiated and stem-like cells. Here, we show that FACT expression is strongly associated with poorly differentiated aggressive cancers with low overall survival. In addition, FACT was found to be upregulated during in vitro transformation and to be necessary, but not sufficient, for driving transformation. FACT also promoted survival and growth of established tumor cells. Genome-wide mapping of chromatin-bound FACT indicated that FACT's role in cancer most likely involves selective chromatin remodeling of genes that stimulate proliferation, inhibit cell death and differentiation, and regulate cellular stress responses.

FUSIM: a software tool for simulating fusion transcripts.

BACKGROUND: Gene fusions are the result of chromosomal aberrations and encode chimeric RNA (fusion transcripts) that play an important role in cancer genesis. Recent advances in high throughput transcriptome sequencing have given rise to computational methods for new fusion discovery. The ability to simulate fusion transcripts is essential for testing and improving those tools. RESULTS: To facilitate this need, we developed FUSIM (FUsion SIMulator), a software tool for simulating fusion transcripts. The simulation of events known to create fusion genes and their resulting chimeric proteins is supported, including inter-chromosome translocation, trans-splicing, complex chromosomal rearrangements, and transcriptional read through events. CONCLUSIONS: FUSIM provides the ability to assemble a dataset of fusion transcripts useful for testing and benchmarking applications in fusion gene discovery.

A new normalizing algorithm for BAC CGH arrays with quality control metrics.

The main focus in pin-tip (or print-tip) microarray analysis is determining which probes, genes, or oligonucleotides are differentially expressed. Specifically in array comparative genomic hybridization (aCGH) experiments, researchers search for chromosomal imbalances in the genome. To model this data, scientists apply statistical methods to the structure of the experiment and assume that the data consist of the signal plus random noise. In this paper we propose "SmoothArray", a new method to preprocess comparative genomic hybridization (CGH) bacterial artificial chromosome (BAC) arrays and we show the effects on a cancer dataset. As part of our R software package "aCGHplus," this freely available algorithm removes the variation due to the intensity effects, pin/print-tip, the spatial location on the microarray chip, and the relative location from the well plate. removal of this variation improves the downstream analysis and subsequent inferences made on the data. Further, we present measures to evaluate the quality of the dataset according to the arrayer pins, 384-well plates, plate rows, and plate columns. We compare our method against competing methods using several metrics to measure the biological signal. With this novel normalization algorithm and quality control measures, the user can improve their inferences on datasets and pinpoint problems that may arise in their BAC aCGH technology.

Comparative survival analysis of breast cancer microarray studies identifies important prognostic genetic pathways.

BACKGROUND: An estimated 12% of females in the United States will develop breast cancer in their lifetime. Although, there are advances in treatment options including surgery and chemotherapy, breast cancer is still the second most lethal cancer in women. Thus, there is a clear need for better methods to predict prognosis for each breast cancer patient. With the advent of large genetic databases and the reduction in cost for the experiments, researchers are faced with choosing from a large pool of potential prognostic markers from numerous breast cancer gene expression profile studies. METHODS: Five microarray datasets related to breast cancer were examined using gene set analysis and the cancers were categorized into different subtypes using a scoring system based on genetic pathway activity. RESULTS: We have observed that significant genes in the individual studies show little reproducibility across the datasets. From our comparative analysis, using gene pathways with clinical variables is more reliable across studies and shows promise in assessing a patient's prognosis. CONCLUSIONS: This study concludes that, in light of clinical variables, there are significant gene pathways in common across the datasets. Specifically, several pathways can further significantly stratify patients for survival. These candidate pathways should help to develop a panel of significant biomarkers for the prognosis of breast cancer patients in a clinical setting.

Error control variability in pathway-based microarray analysis.

MOTIVATION: The decision to commit some or many false positives in practice rests with the investigator. Unfortunately, not all error control procedures perform the same. Our problem is to choose an error control procedure to determine a P-value threshold for identifying differentially expressed pathways in high-throughput gene expression studies. Pathway analysis involves fewer tests than differential gene expression analysis, on the order of a few hundred. We discuss and compare methods for error control for pathway analysis with gene expression data. RESULTS: In consideration of the variability in test results, we find that the widely used Benjamini and Hochberg's (BH) false discovery rate (FDR) analysis is less robust than alternative procedures. BH's error control requires a large number of hypothesis tests, a reasonable assumption for differential gene expression analysis, though not the case with pathway-based analysis. Therefore, we advocate through a series of simulations and applications to real gene expression data that researchers control the number of false positives rather than the FDR.