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.

Flype: Software for enabling personalized medicine.

The advent of next generation DNA sequencing (NGS) has revolutionized clinical medicine by enabling wide-spread testing for genomic anomalies and polymorphisms. With that explosion in testing, however, come several informatics challenges including managing large amounts of data, interpreting the results and providing clinical decision support. We present Flype, a web-based bioinformatics platform built by a small group of bioinformaticians working in a community hospital setting, to address these challenges by allowing us to: (a) securely accept data from a variety of sources, (b) send orders to a variety of destinations, (c) perform secondary analysis and annotation of NGS data, (d) provide a central repository for all genomic variants, (e) assist with tertiary analysis and clinical interpretation, (f) send signed out data to our EHR as both PDF and discrete data elements, (g) allow population frequency analysis and (h) update variant annotation when literature knowledge evolves. We discuss the multiple use cases Flype supports such as (a) in-house NGS tests, (b) in-house pharmacogenomics (PGX) tests, (c) dramatic scale-up of genomic testing using an external lab, (d) consumer genomics using two external partners, and (e) a variety of reporting tools. The source code for Flype is available upon request to the authors.

LyP-1-Modified Oncolytic Adenoviruses Targeting Transforming Growth Factor ß Inhibit Tumor Growth and Metastases and Augment Immune Checkpoint Inhibitor Therapy in Breast Cancer Mouse Models.

We report here the development of oncolytic adenoviruses (Ads) that have reduced toxicity, enhanced tumor tropism, produce strong antitumor response, and can overcome resistance to immune checkpoint inhibitor therapy in breast cancer. We have shown that LyP-1 receptor (p32) is highly expressed on the surface of breast cancer cells and tumors from cancer patients, and that increased stromal expression of transforming growth factor ß-1 (TGFß-1) is associated with triple-negative breast cancer. Therefore, we constructed oncolytic Ads, AdLyp.sT and mHAdLyp.sT, in which the p32-binding LyP-1 peptide was genetically inserted into the adenoviral fiber protein. Both AdLyp.sT and mHAdLyp.sT express sTGFßRIIFc, a TGFß decoy that can inhibit TGFß pathways. mHAdLyp.sT is an Ad5/48 chimeric hexon virus in which hypervariable regions (HVRs 1-7) of Ad5 are replaced with the corresponding Ad48 HVRs. AdLyp.sT and mHAdLyp.sT exhibited better binding, replication, and produced higher sTGFßRIIFc protein levels in breast cancer cell lines compared with Ad.sT or mHAd.sT control viruses without LyP-1 peptide modification. Systemic delivery of mHAdLyp.sT in mice resulted in reduced hepatic/systemic toxicity compared with Ad.sT and AdLyp.sT. Intravenous delivery of AdLyp.sT and mHAdLyp.sT elicited a strong antitumor response in a human MDA-MB-231 bone metastasis model in mice, as indicated by bioluminescence imaging, radiographic tumor burden, serum TRACP 5b and calcium, and body weight analyses. Furthermore, intratumoral delivery of AdLyp.sT in 4T1 model in immunocompetent mice inhibited tumor growth and metastases, and augmented anti-PD-1 and anti-CTLA-4 therapy. Based on these studies, we believe that AdLyp.sT and mHAdLyp.sT can be developed as potential targeted immunotherapy agents for the treatment of breast cancer.

An Oncolytic Adenovirus Targeting Transforming Growth Factor ß Inhibits Protumorigenic Signals and Produces Immune Activation: A Novel Approach to Enhance Anti-PD-1 and Anti-CTLA-4 Therapy.

In an effort to develop a new therapy for cancer and to improve antiprogrammed death inhibitor-1 (anti-PD-1) and anticytotoxic T lymphocyte-associated protein (anti-CTLA-4) responses, we have created a telomerase reverse transcriptase promoter-regulated oncolytic adenovirus rAd.sT containing a soluble transforming growth factor receptor II fused with human IgG Fc fragment (sTGFßRIIFc) gene. Infection of breast and renal tumor cells with rAd.sT produced sTGFßRIIFc protein with dose-dependent cytotoxicity. In immunocompetent mouse 4T1 breast tumor model, intratumoral delivery of rAd.sT inhibited both tumor growth and lung metastases. rAd.sT downregulated the expression of several transforming growth factor ß (TGFß) target genes involved in tumor growth and metastases, inhibited Th2 cytokine expression, and induced Th1 cytokines and chemokines, and granzyme B and perforin expression. rAd.sT treatment also increased the percentage of CD8+ T lymphocytes, promoted the generation of CD4+ T memory cells, reduced regulatory T lymphocytes (Tregs), and reduced bone marrow-derived suppressor cells. Importantly, rAd.sT treatment increased the percentage of CD4+ T lymphocytes, and promoted differentiation and maturation of antigen-presenting dendritic cells in the spleen. In the immunocompetent mouse Renca renal tumor model, similar therapeutic effects and immune activation results were observed. In the 4T1 mammary tumor model, rAd.sT improved the inhibition of tumor growth and lung and liver metastases by anti-PD-1 and anti-CTLA-4 antibodies. Analysis of the human breast and kidney tumors showed that a significant number of tumor tissues expressed high levels of TGFß and TGFß-inducible genes. Therefore, rAd.sT could be a potential enhancer of anti-PD-1 and anti-CTLA-4 therapy for treating breast and kidney cancers.

Germline Mutations in ATM and BRCA1/2 Distinguish Risk for Lethal and Indolent Prostate Cancer and are Associated with Early Age at Death.

BACKGROUND: Germline mutations in BRCA1/2 and ATM have been associated with prostate cancer (PCa) risk. OBJECTIVE: To directly assess whether germline mutations in these three genes distinguish lethal from indolent PCa and whether they confer any effect on age at death. DESIGN, SETTING, AND PARTICIPANTS: A retrospective case-case study of 313 patients who died of PCa and 486 patients with low-risk localized PCa of European, African, and Chinese descent. Germline DNA of each of the 799 patients was sequenced for these three genes. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Mutation carrier rates and their effect on lethal PCa were analyzed using the Fisher's exact test and Cox regression analysis, respectively. RESULTS AND LIMITATIONS: The combined BRCA1/2 and ATM mutation carrier rate was significantly higher in lethal PCa patients (6.07%) than localized PCa patients (1.44%), p=0.0007. The rate also differed significantly among lethal PCa patients as a function of age at death (10.00%, 9.08%, 8.33%, 4.94%, and 2.97% in patients who died ≤ 60 yr, 61-65 yr, 66-70 yr, 71-75 yr, and over 75 yr, respectively, p=0.046) and time to death after diagnosis (12.26%, 4.76%, and 0.98% in patients who died ≤ 5 yr, 6-10 yr, and>10 yr after a PCa diagnosis, respectively, p=0.0006). Survival analysis in the entire cohort revealed mutation carriers remained an independent predictor of lethal PCa after adjusting for race and age, prostate-specific antigen, and Gleason score at the time of diagnosis (hazard ratio=2.13, 95% confidence interval: 1.24-3.66, p=0.004). A limitation of this study is that other DNA repair genes were not analyzed. CONCLUSIONS: Mutation status of BRCA1/2 and ATM distinguishes risk for lethal and indolent PCa and is associated with earlier age at death and shorter survival time. PATIENT SUMMARY: Prostate cancer patients with inherited mutations in BRCA1/2 and ATM are more likely to die of prostate cancer and do so at an earlier age.

Ultra-fast local-haplotype variant calling using paired-end DNA-sequencing data reveals somatic mosaicism in tumor and normal blood samples.

Somatic mosaicism refers to the existence of somatic mutations in a fraction of somatic cells in a single biological sample. Its importance has mainly been discussed in theory although experimental work has started to emerge linking somatic mosaicism to disease diagnosis. Through novel statistical modeling of paired-end DNA-sequencing data using blood-derived DNA from healthy donors as well as DNA from tumor samples, we present an ultra-fast computational pipeline, LocHap that searches for multiple single nucleotide variants (SNVs) that are scaffolded by the same reads. We refer to scaffolded SNVs as local haplotypes (LH). When an LH exhibits more than two genotypes, we call it a local haplotype variant (LHV). The presence of LHVs is considered evidence of somatic mosaicism because a genetically homogeneous cell population will not harbor LHVs. Applying LocHap to whole-genome and whole-exome sequence data in DNA from normal blood and tumor samples, we find wide-spread LHVs across the genome. Importantly, we find more LHVs in tumor samples than in normal samples, and more in older adults than in younger ones. We confirm the existence of LHVs and somatic mosaicism by validation studies in normal blood samples. LocHap is publicly available at http://www.compgenome.org/lochap.

Bayesian inference for intratumour heterogeneity in mutations and copy number variation.

Tumor samples are heterogeneous. They consist of different subclones that are characterized by differences in DNA nucleotide sequences and copy numbers on multiple loci. Heterogeneity can be measured through the identification of the subclonal copy number and sequence at a selected set of loci. Understanding that the accurate identification of variant allele fractions greatly depends on a precise determination of copy numbers, we develop a Bayesian feature allocation model for jointly calling subclonal copy numbers and the corresponding allele sequences for the same loci. The proposed method utilizes three random matrices, L , Z and w to represent subclonal copy numbers ( L ), numbers of subclonal variant alleles ( Z ) and cellular fractions of subclones in samples ( w ), respectively. The unknown number of subclones implies a random number of columns for these matrices. We use next-generation sequencing data to estimate the subclonal structures through inference on these three matrices. Using simulation studies and a real data analysis, we demonstrate how posterior inference on the subclonal structure is enhanced with the joint modeling of both structure and sequencing variants on subclonal genomes. Software is available at http://compgenome.org/BayClone2.

MAD Bayes for Tumor Heterogeneity - Feature Allocation with Exponential Family Sampling.

We propose small-variance asymptotic approximations for inference on tumor heterogeneity (TH) using next-generation sequencing data. Understanding TH is an important and open research problem in biology. The lack of appropriate statistical inference is a critical gap in existing methods that the proposed approach aims to fill. We build on a hierarchical model with an exponential family likelihood and a feature allocation prior. The proposed implementation of posterior inference generalizes similar small-variance approximations proposed by Kulis and Jordan (2012) and Broderick et.al (2012b) for inference with Dirichlet process mixture and Indian buffet process prior models under normal sampling. We show that the new algorithm can successfully recover latent structures of different haplotypes and subclones and is magnitudes faster than available Markov chain Monte Carlo samplers. The latter are practically infeasible for high-dimensional genomics data. The proposed approach is scalable, easy to implement and benefits from the exibility of Bayesian nonparametric models. More importantly, it provides a useful tool for applied scientists to estimate cell subtypes in tumor samples. R code is available on http://www.ma.utexas.edu/users/yxu/.

Zodiac: A Comprehensive Depiction of Genetic Interactions in Cancer by Integrating TCGA Data.

BACKGROUND: Genetic interactions play a critical role in cancer development. Existing knowledge about cancer genetic interactions is incomplete, especially lacking evidences derived from large-scale cancer genomics data. The Cancer Genome Atlas (TCGA) produces multimodal measurements across genomics and features of thousands of tumors, which provide an unprecedented opportunity to investigate the interplays of genes in cancer. METHODS: We introduce Zodiac, a computational tool and resource to integrate existing knowledge about cancer genetic interactions with new information contained in TCGA data. It is an evolution of existing knowledge by treating it as a prior graph, integrating it with a likelihood model derived by Bayesian graphical model based on TCGA data, and producing a posterior graph as updated and data-enhanced knowledge. In short, Zodiac realizes "Prior interaction map + TCGA data → Posterior interaction map." RESULTS: Zodiac provides molecular interactions for about 200 million pairs of genes. All the results are generated from a big-data analysis and organized into a comprehensive database allowing customized search. In addition, Zodiac provides data processing and analysis tools that allow users to customize the prior networks and update the genetic pathways of their interest. Zodiac is publicly available at www.compgenome.org/ZODIAC. CONCLUSIONS: Zodiac recapitulates and extends existing knowledge of molecular interactions in cancer. It can be used to explore novel gene-gene interactions, transcriptional regulation, and other types of molecular interplays in cancer.

Bayclone: Bayesian nonparametric inference of tumor subclones using NGS data.

In this paper, we present a novel feature allocation model to describe tumor heterogeneity (TH) using next-generation sequencing (NGS) data. Taking a Bayesian approach, we extend the Indian buffet process (IBP) to define a class of nonparametric models, the categorical IBP (cIBP). A cIBP takes categorical values to denote homozygous or heterozygous genotypes at each SNV. We define a subclone as a vector of these categorical values, each corresponding to an SNV. Instead of partitioning somatic mutations into non-overlapping clusters with similar cellular prevalences, we took a different approach using feature allocation. Importantly, we do not assume somatic mutations with similar cellular prevalence must be from the same subclone and allow overlapping mutations shared across subclones. We argue that this is closer to the underlying theory of phylogenetic clonal expansion, as somatic mutations occurred in parent subclones should be shared across the parent and child subclones. Bayesian inference yields posterior probabilities of the number, genotypes, and proportions of subclones in a tumor sample, thereby providing point estimates as well as variabilities of the estimates for each subclone. We report results on both simulated and real data. BayClone is available at http://health.bsd.uchicago.edu/yji/soft.html.

Direct observation of genomic heterogeneity through local haplotyping analysis.

BACKGROUND: It has been an abiding belief among geneticists that multicellular organisms' genomes can be analyzed under the assumption that a single individual has a uniform genome in all its cells. Despite some evidence to the contrary, this belief has been used as an axiomatic assumption in most genome analysis software packages. In this paper we present observations in human whole genome data, human whole exome data and in mouse whole genome data to challenge this assumption. We show that heterogeneity is in fact ubiquitous and readily observable in ordinary Next Generation Sequencing (NGS) data. RESULTS: Starting with the assumption that a single NGS read (or read pair) must come from one haplotype, we built a procedure for directly observing haplotypes at a local level by examining 2 or 3 adjacent single nucleotide polymorphisms (SNPs) which are close enough on the genome to be spanned by individual reads. We applied this procedure to NGS data from three different sources: whole genome of a Central European trio from the 1000 genomes project, whole genome data from laboratory-bred strains of mouse, and whole exome data from a set of patients of head and neck tumors. Thousands of loci were found in each genome where reads spanning 2 or 3 SNPs displayed more than two haplotypes, indicating that the locus is heterogeneous. We show that such loci are ubiquitous in the genome and cannot be explained by segmental duplications. We explain them on the basis of cellular heterogeneity at the genomic level. Such heterogeneous loci were found in all normal and tumor genomes examined. CONCLUSIONS: Our results highlight the need for new methods to analyze genomic variation because existing ones do not systematically consider local haplotypes. Identification of cancer somatic mutations is complicated because of tumor heterogeneity. It is further complicated if, as we show, normal tissues are also heterogeneous. Methods for biomarker discovery must consider contextual haplotype information rather than just whether a variant "is present".

Predicting GPCR-G-protein coupling using hidden Markov models.

MOTIVATION: Determining the coupling specificity of G-protein coupled receptors (GPCRs) is important for understanding the biology of this class of pharmacologically important proteins. Currently available in silico methods for predicting GPCR-G-protein coupling specificity have high error rate. METHOD: We introduce a new approach for creating hidden Markov models (HMMs) based on a first guess about the importance of various residues. We call these knowledge restricted HMMs to emphasize the fact that the state space of the HMM is restricted by the application of a priori knowledge. Specifically, we use only those amino acid residues of GPCRs which are likely to interact with G-proteins, namely those that are predicted to be in the intra-cellular loops. Furthermore, we concatenate these predicted loops into one sequence rather than considering them as four disparate units. This reduces the HMM state space by drastically decreasing the sequence length. RESULTS: Our knowledge restricted HMM based method to predict GPCR-G-protein coupling specificity has an error rate of <1%, when applied to a test set of GPCRs with known G-protein coupling specificity. AVAILABILITY: Academic users can get the data set mentioned herein and HMMs from the authors.

Immunoinformatics in personalized medicine.

Diagnosis of human disease has been undergoing steady improvement over the past few centuries. Many ailments that were once considered a single entity have been classified into finer categories on the basis of response to therapy (e.g. type I and type II diabetes), inheritance (e.g. familial and non-familial polyposis coli), histology (e.g. small cell and adenocarcinoma of lung) and most recently transcriptional profiling (e.g. leukaemia, lymphoma). The next dimension in this finer categorization appears to be the typing of the patient rather than the disease i.e. disease X in person of type Y. The problem of personalized medicine is to devise tests which predict the type of individual, especially where the type is correlated with response to therapy. Immunology has been at the forefront of personalized medicine for quite a while, even though the term is not often used in this connection. Blood grouping and cross-matching (for blood transfusion), and anaphylaxis test (for penicillin) are just two examples. In this paper I will argue that immunological tests have an important place in the future of personalized medicine. I will describe methods we developed for personalizing vaccines based on MHC allele frequencies in human populations and methods for predicting peptide binding to class I MHC molecules. In conclusion, I will argue that immunological tests, and consequently immunoinformatics, will play a big role in making personalized medicine a reality.

Characterization of mGluR5R, a novel, metabotropic glutamate receptor 5-related gene.

We report here the isolation of a novel gene termed mGluR5R (mGluR5-related). The N-terminus of mGluR5R is highly similar to the extracellular domain of metabotropic glutamate receptor 5 (mGluR5) whereas the C-terminus bears similarity to the testis-specific gene, RNF18. mGluR5R is expressed in the human CNS in a coordinate fashion with mGluR5. Although the sequence suggests that mGluR5R may be a secreted glutamate binding protein, we found that when expressed in HEK293 cells it was membrane associated and not secreted. Furthermore, mGluR5R was incapable of binding the metabotropic glutamate receptor class I selective agonist, quisqualate. Although mGluR5R could not form disulfide-mediated covalent homodimers, it was able to form a homomeric complex, presumably through noncovalent interactions. mGluR5R also formed noncovalent heteromeric associations with an engineered construct of the extracellular domain of mGluR5 as well as with full-length mGluR5 and mGluR1alpha. The ability of mGluR5R to associate with mGluR1alpha and mGluR5 suggests that it may be a modulator of class I metabotropic glutamate receptor function.