FuncPEP v2.0: An Updated Database of Functional Short Peptides Translated from Non-Coding RNAs

Over the past decade, there have been reports of short novel functional peptides (less than 100 aa in length) translated from so-called non-coding RNAs (ncRNAs) that have been characterized using mass spectrometry (MS) and large-scale proteomics studies. Therefore, understanding the bivalent functions of some ncRNAs as transcripts that encode both functional RNAs and short peptides, which we named ncPEPs, will deepen our understanding of biology and disease. In 2020, we published the first database of functional peptides translated from non-coding RNAs-FuncPEP. Herein, we have performed an update including the newly published ncPEPs from the last 3 years along with the categorization of host ncRNAs. FuncPEP v2.0 contains 152 functional ncPEPs, out of which 40 are novel entries. A PubMed search from August 2020 to July 2023 incorporating specific keywords was performed and screened for publications reporting validated functional peptides derived from ncRNAs. We did not observe a significant increase in newly discovered functional ncPEPs, but a steady increase. The novel identified ncPEPs included in the database were characterized by a wide array of molecular and physiological parameters (i.e., types of host ncRNA, species distribution, chromosomal density, distribution of ncRNA length, identification methods, molecular weight, and functional distribution across humans and other species). We consider that, despite the fact that MS can now easily identify ncPEPs, there still are important limitations in proving their functionality.

RETRACTED: Liu et al. Prediction of Ovarian Cancer Response to Therapy Based on Deep Learning Analysis of Histopathology Images. Cancers 2023, 15, 4044

The journal and authors wish to retract the article entitled 'Prediction of Ovarian Cancer Response to Therapy Based on Deep Learning Analysis of Histopathology Images' cited above [...].

PCA-Plus: Enhanced principal component analysis with illustrative applications to batch effects and their quantitation

Background

Principal component analysis (PCA), a standard approach to analysis and visualization of large datasets, is commonly used in biomedical research for detecting similarities and differences among groups of samples. We initially used conventional PCA as a tool for critical quality control of batch and trend effects in multi-omic profiling data produced by The Cancer Genome Atlas (TCGA) project of the NCI. We found, however, that conventional PCA visualizations were often hard to interpret when inter-batch differences were moderate in comparison with intra-batch differences; it was also difficult to quantify batch effects objectively. We, therefore, sought enhancements to make the method more informative in those and analogous settings.

Results

We have developed algorithms and a toolbox of enhancements to conventional PCA that improve the detection, diagnosis, and quantitation of differences between or among groups, e.g., groups of molecularly profiled biological samples. The enhancements include (i) computed group centroids; (ii) sample-dispersion rays; (iii) differential coloring of centroids, rays, and sample data points; (iii) trend trajectories; and (iv) a novel separation index (DSC) for quantitation of differences among groups.

Conclusions

PCA-Plus has been our most useful single tool for analyzing, visualizing, and quantitating batch effects, trend effects, and class differences in molecular profiling data of many types: mRNA expression, microRNA expression, DNA methylation, and DNA copy number. An early version of PCA-Plus has been used as the central graphical visualization in our MBatch package for near-real-time surveillance of data for analysis working groups in more than 70 TCGA, PanCancer Atlas, PanCancer Analysis of Whole Genomes, and Genome Data Analysis Network projects of the NCI. The algorithms and software are generic, hence applicable more generally to other types of multivariate data as well. PCA-Plus is freely available in a down-loadable R package at our MBatch website.

SPOP Mutations Target STING1 Signaling in Prostate Cancer and Create Therapeutic Vulnerabilities to PARP Inhibitor-Induced Growth Suppression

PURPOSE

Speckle-type POZ protein (SPOP) is important in DNA damage response (DDR) and maintenance of genomic stability. Somatic heterozygous missense mutations in the SPOP substrate-binding cleft are found in up to 15% of prostate cancers. While mutations in SPOP predict for benefit from androgen receptor signaling inhibition (ARSi) therapy, outcomes for patients with SPOP-mutant (SPOPmut) prostate cancer are heterogeneous and targeted treatments for SPOPmut castrate-resistant prostate cancer (CRPC) are lacking.

EXPERIMENTAL DESIGN

Using in silico genomic and transcriptomic tumor data, proteomics analysis, and genetically modified cell line models, we demonstrate mechanistic links between SPOP mutations, STING signaling alterations, and PARP inhibitor vulnerabilities.

RESULTS

We demonstrate that SPOP mutations are associated with upregulation of a 29-gene noncanonical (NC) STING (NC-STING) signature in a subset of SPOPmut, treatment-refractory CRPC patients. We show in preclinical CRPC models that SPOP targets and destabilizes STING1 protein, and prostate cancer-associated SPOP mutations result in upregulated NC-STING-NF-κB signaling and macrophage- and tumor microenvironment (TME)-facilitated reprogramming, leading to tumor cell growth. Importantly, we provide in vitro and in vivo mechanism-based evidence that PARP inhibitor (PARPi) treatment results in a shift from immunosuppressive NC-STING-NF-κB signaling to antitumor, canonical cGAS-STING-IFNβ signaling in SPOPmut CRPC and results in enhanced tumor growth inhibition.

CONCLUSIONS

We provide evidence that SPOP is critical in regulating immunosuppressive versus antitumor activity downstream of DNA damage-induced STING1 activation in prostate cancer. PARPi treatment of SPOPmut CRPC alters this NC-STING signaling toward canonical, antitumor cGAS-STING-IFNβ signaling, highlighting a novel biomarker-informed treatment strategy for prostate cancer.

Prediction of Ovarian Cancer Response to Therapy Based on Deep Learning Analysis of Histopathology Images

BACKGROUND

Ovarian cancer remains the leading gynecological cause of cancer mortality. Predicting the sensitivity of ovarian cancer to chemotherapy at the time of pathological diagnosis is a goal of precision medicine research that we have addressed in this study using a novel deep-learning neural network framework to analyze the histopathological images.

METHODS

We have developed a method based on the Inception V3 deep learning algorithm that complements other methods for predicting response to standard platinum-based therapy of the disease. For the study, we used histopathological H&E images (pre-treatment) of high-grade serous carcinoma from The Cancer Genome Atlas (TCGA) Genomic Data Commons portal to train the Inception V3 convolutional neural network system to predict whether cancers had independently been labeled as sensitive or resistant to subsequent platinum-based chemotherapy. The trained model was then tested using data from patients left out of the training process. We used receiver operating characteristic (ROC) and confusion matrix analyses to evaluate model performance and Kaplan-Meier survival analysis to correlate the predicted probability of resistance with patient outcome. Finally, occlusion sensitivity analysis was piloted as a start toward correlating histopathological features with a response.

RESULTS

The study dataset consisted of 248 patients with stage 2 to 4 serous ovarian cancer. For a held-out test set of forty patients, the trained deep learning network model distinguished sensitive from resistant cancers with an area under the curve (AUC) of 0.846 ± 0.009 (SE). The probability of resistance calculated from the deep-learning network was also significantly correlated with patient survival and progression-free survival. In confusion matrix analysis, the network classifier achieved an overall predictive accuracy of 85% with a sensitivity of 73% and specificity of 90% for this cohort based on the Youden-J cut-off. Stage, grade, and patient age were not statistically significant for this cohort size. Occlusion sensitivity analysis suggested histopathological features learned by the network that may be associated with sensitivity or resistance to the chemotherapy, but multiple marker studies will be necessary to follow up on those preliminary results.

CONCLUSIONS

This type of analysis has the potential, if further developed, to improve the prediction of response to therapy of high-grade serous ovarian cancer and perhaps be useful as a factor in deciding between platinum-based and other therapies. More broadly, it may increase our understanding of the histopathological variables that predict response and may be adaptable to other cancer types and imaging modalities.

RPPA SPACE: an R package for normalization and quantitation of Reverse-Phase Protein Array data

SUMMARY

Reverse-Phase Protein Array (RPPA) is a robust high-throughput, cost-effective platform for quantitatively measuring proteins in biological specimens. However, converting raw RPPA data into normalized, analysis-ready data remains a challenging task. Here, we present the RPPA SPACE (RPPA Superposition Analysis and Concentration Evaluation) R package, a substantially improved successor to SuperCurve, to meet that challenge. SuperCurve has been used to normalize over 170 000 samples to date. RPPA SPACE allows exclusion of poor-quality samples from the normalization process to improve the quality of the remaining samples. It also features a novel quality-control metric, 'noise', that estimates the level of random errors present in each RPPA slide. The noise metric can help to determine the quality and reliability of the data. In addition, RPPA SPACE has simpler input requirements and is more flexible than SuperCurve, it is much faster with greatly improved error reporting.

AVAILABILITY AND IMPLEMENTATION

The standalone RPPA SPACE R package, tutorials and sample data are available via https://rppa.space/, CRAN (https://cran.r-project.org/web/packages/RPPASPACE/index.html) and GitHub (https://github.com/MD-Anderson-Bioinformatics/RPPASPACE).

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Fibroblast Growth Factor Receptor 1 Drives the Metastatic Progression of Prostate Cancer

BACKGROUND

No curative therapy is currently available for metastatic prostate cancer (PCa). The diverse mechanisms of progression include fibroblast growth factor (FGF) axis activation.

OBJECTIVE

To investigate the molecular and clinical implications of fibroblast growth factor receptor 1 (FGFR1) and its isoforms (α/β) in the pathogenesis of PCa bone metastases.

DESIGN, SETTING, AND PARTICIPANTS

In silico, in vitro, and in vivo preclinical approaches were used. RNA-sequencing and immunohistochemical (IHC) studies in human samples were conducted.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS

In mice, bone metastases (chi-square/Fisher's test) and survival (Mantel-Cox) were assessed. In human samples, FGFR1 and ladinin 1 (LAD1) analysis associated with PCa progression were evaluated (IHC studies, Fisher's test).

RESULTS AND LIMITATIONS

FGFR1 isoform expression varied among PCa subtypes. Intracardiac injection of mice with FGFR1-expressing PC3 cells reduced mouse survival (α, p < 0.0001; β, p = 0.032) and increased the incidence of bone metastases (α, p < 0.0001; β, p = 0.02). Accordingly, IHC studies of human castration-resistant PCa (CRPC) bone metastases revealed significant enrichment of FGFR1 expression compared with treatment-naïve, nonmetastatic primary tumors (p = 0.0007). Expression of anchoring filament protein LAD1 increased in FGFR1-expressing PC3 cells and was enriched in human CRPC bone metastases (p = 0.005).

CONCLUSIONS

FGFR1 expression induces bone metastases experimentally and is significantly enriched in human CRPC bone metastases, supporting its prometastatic effect in PCa. LAD1 expression, found in the prometastatic PCa cells expressing FGFR1, was also enriched in CRPC bone metastases. Our studies support and provide a roadmap for the development of FGFR blockade for advanced PCa.

PATIENT SUMMARY

We studied the role of fibroblast growth factor receptor 1 (FGFR1) in prostate cancer (PCa) progression. We found that PCa cells with high FGFR1 expression increase metastases and that FGFR1 expression is increased in human PCa bone metastases, and identified genes that could participate in the metastases induced by FGFR1. These studies will help pinpoint PCa patients who use fibroblast growth factor to progress and will benefit by the inhibition of this pathway.

PARP and CDK4/6 Inhibitor Combination Therapy Induces Apoptosis and Suppresses Neuroendocrine Differentiation in Prostate Cancer

We analyzed the efficacy and mechanistic interactions of PARP inhibition (PARPi; olaparib) and CDK4/6 inhibition (CDK4/6i; palbociclib or abemaciclib) combination therapy in castration-resistant prostate cancer (CRPC) and neuroendocrine prostate cancer (NEPC) models. We demonstrated that combined olaparib and palbociblib or abemaciclib treatment resulted in synergistic suppression of the p-Rb1-E2F1 signaling axis at the transcriptional and posttranslational levels, leading to disruption of cell-cycle progression and inhibition of E2F1 gene targets, including genes involved in DDR signaling/damage repair, antiapoptotic BCL-2 family members (BCL-2 and MCL-1), CDK1, and neuroendocrine differentiation (NED) markers in vitro and in vivo In addition, olaparib + palbociclib or olaparib + abemaciclib combination treatment resulted in significantly greater growth inhibition and apoptosis than either single agent alone. We further showed that PARPi and CDK4/6i combination treatment-induced CDK1 inhibition suppressed p-S70-BCL-2 and increased caspase cleavage, while CDK1 overexpression effectively prevented the downregulation of p-S70-BCL-2 and largely rescued the combination treatment-induced cytotoxicity. Our study defines a novel combination treatment strategy for CRPC and NEPC and demonstrates that combination PARPi and CDK4/6i synergistically promotes suppression of the p-Rb1-E2F1 axis and E2F1 target genes, including CDK1 and NED proteins, leading to growth inhibition and increased apoptosis in vitro and in vivo Taken together, our results provide a molecular rationale for PARPi and CDK4/6i combination therapy and reveal mechanism-based clinical trial opportunities for men with NEPC.

ATR Inhibition Induces CDK1-SPOP Signaling and Enhances Anti-PD-L1 Cytotoxicity in Prostate Cancer

PURPOSE

Despite significant benefit for other cancer subtypes, immune checkpoint blockade (ICB) therapy has not yet been shown to significantly improve outcomes for men with castration-resistant prostate cancer (CRPC). Prior data have shown that DNA damage response (DDR) deficiency, via genetic alteration and/or pharmacologic induction using DDR inhibitors (DDRi), may improve ICB response in solid tumors in part due to induction of mitotic catastrophe and innate immune activation. Discerning the underlying mechanisms of this DDRi-ICB interaction in a prostate cancer-specific manner is vital to guide novel clinical trials and provide durable clinical responses for men with CRPC.

EXPERIMENTAL DESIGN

We treated prostate cancer cell lines with potent, specific inhibitors of ATR kinase, as well as with PARP inhibitor, olaparib. We performed analyses of cGAS-STING and DDR signaling in treated cells, and treated a syngeneic androgen-indifferent, prostate cancer model with combined ATR inhibition and anti-programmed death ligand 1 (anti-PD-L1), and performed single-cell RNA sequencing analysis in treated tumors.

RESULTS

ATR inhibitor (ATRi; BAY1895433) directly repressed ATR-CHK1 signaling, activated CDK1-SPOP axis, leading to destabilization of PD-L1 protein. These effects of ATRi are distinct from those of olaparib, and resulted in a cGAS-STING-initiated, IFN-β-mediated, autocrine, apoptotic response in CRPC. The combination of ATRi with anti-PD-L1 therapy resulted in robust innate immune activation and a synergistic, T-cell-dependent therapeutic response in our syngeneic mouse model.

CONCLUSIONS

This work provides a molecular mechanistic rationale for combining ATR-targeted agents with immune checkpoint blockade for patients with CRPC. Multiple early-phase clinical trials of this combination are underway.

Methylation-eQTL Analysis in Cancer Research

MOTIVATION

DNA methylation is a key epigenetic factor regulating gene expression. While promoter methylation has been well studied, recent publications have revealed that functionally important methylation also occurs in intergenic and distal regions, and varies across genes and tissue types. Given the growing importance of inter-platform integrative genomic analyses, there is an urgent need to develop methods to discover and characterize gene-level relationships between methylation and expression.

RESULTS

We introduce a novel sequential penalized regression approach to identify methylation-expression quantitative trait loci (methyl-eQTLs), a term that we have coined to represent, for each gene and tissue type, a sparse set of CpG loci best explaining gene expression and accompanying weights indicating direction and strength of association. Using TCGA and MD Anderson colorectal cohorts to build and validate our models, we demonstrate our strategy better explains expression variability than current commonly used gene-level methylation summaries. The methyl-eQTLs identified by our approach can be used to construct gene-level methylation summaries that are maximally correlated with gene expression for use in integrative models, and produce a tissue-specific summary of which genes appear to be strongly regulated by methylation. Our results introduce an important resource to the biomedical community for integrative genomics analyses involving DNA methylation.

AVAILABILITY AND IMPLEMENTATION

We produce an R Shiny app (https://rstudio-prd-c1.pmacs.upenn.edu/methyl-eQTL/) that interactively presents methyl-eQTL results for colorectal, breast, and pancreatic cancer. The source R code for this work is provided in the supplement.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Downregulation of 15-hydroxyprostaglandin dehydrogenase during acquired tamoxifen resistance and association with poor prognosis in ERα-positive breast cancer

Aim:

Tamoxifen (TAM) resistance remains a clinical issue in breast cancer. The authors previously reported that 15-hydroxyprostaglandin dehydrogenase (HPGD) was significantly downregulated in tamoxifen-resistant (TAMr) breast cancer cell lines. Here, the authors investigated the relationship between HPGD expression, TAM resistance and prediction of outcome in breast cancer.

Methods:

HPGD overexpression and silencing studies were performed in isogenic TAMr and parental human breast cancer cell lines to establish the impact of HPGD expression on TAM resistance. HPGD expression and clinical outcome relationships were explored using immunohistochemistry and in silico analysis.

Results:

Restoration of HPGD expression and activity sensitised TAMr MCF-7 cells to TAM and 17β-oestradiol, whilst HPGD silencing in parental MCF-7 cells reduced TAM sensitivity. TAMr cells released more prostaglandin E₂ (PGE₂) than controls, which was reduced in TAMr cells stably transfected with HPGD. Exogenous PGE₂ signalled through the EP4 receptor to reduce breast cancer cell sensitivity to TAM. Decreased HPGD expression was associated with decreased overall survival in ERα-positive breast cancer patients.

Conclusions:

HPGD downregulation in breast cancer is associated with reduced response to TAM therapy via PGE₂-EP4 signalling and decreases patient survival. The data offer a potential target to develop combination therapies that may overcome acquired tamoxifen resistance.

Development and Validation of a Gene Signature Classifier for Consensus Molecular Subtyping of Colorectal Carcinoma in a CLIA-Certified Setting

PURPOSE

Consensus molecular subtyping (CMS) of colorectal cancer has potential to reshape the colorectal cancer landscape. We developed and validated an assay that is applicable on formalin-fixed, paraffin-embedded (FFPE) samples of colorectal cancer and implemented the assay in a Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory.

EXPERIMENTAL DESIGN

We performed an in silico experiment to build an optimal CMS classifier using a training set of 1,329 samples from 12 studies and validation set of 1,329 samples from 14 studies. We constructed an assay on the basis of NanoString CodeSets for the top 472 genes, and performed analyses on paired flash-frozen (FF)/FFPE samples from 175 colorectal cancers to adapt the classifier to FFPE samples using a subset of genes found to be concordant between FF and FFPE, tested the classifier's reproducibility and repeatability, and validated in a CLIA-certified laboratory. We assessed prognostic significance of CMS in 345 patients pooled across three clinical trials.

RESULTS

The best classifier was weighted support vector machine with high accuracy across platforms and gene lists (>0.95), and the 472-gene model outperforming existing classifiers. We constructed subsets of 99 and 200 genes with high FF/FFPE concordance, and adapted FFPE-based classifier that had strong classification accuracy (>80%) relative to "gold standard" CMS. The classifier was reproducible to sample type and RNA quality, and demonstrated poor prognosis for CMS1-3 and good prognosis for CMS2 in metastatic colorectal cancer (P < 0.001).

CONCLUSIONS

We developed and validated a colorectal cancer CMS assay that is ready for use in clinical trials, to assess prognosis in standard-of-care settings and explore as predictor of therapy response.

FuncPEP: A Database of Functional Peptides Encoded by Non-Coding RNAs

Non-coding RNAs (ncRNAs) are essential players in many cellular processes, from normal development to oncogenic transformation. Initially, ncRNAs were defined as transcripts that lacked an open reading frame (ORF). However, multiple lines of evidence suggest that certain ncRNAs encode small peptides of less than 100 amino acids. The sequences encoding these peptides are known as small open reading frames (smORFs), many initiating with the traditional AUG start codon but terminating with atypical stop codons, suggesting a different biogenesis. The ncRNA-encoded peptides (ncPEPs) are gradually becoming appreciated as a new class of functional molecules that contribute to diverse cellular processes, and are deregulated in different diseases contributing to pathogenesis. As multiple publications have identified unique ncPEPs, we appreciated the need for assembling a new web resource that could gather information about these functional ncPEPs. We developed FuncPEP, a new database of functional ncRNA encoded peptides, containing all experimentally validated and functionally characterized ncPEPs. Currently, FuncPEP includes a comprehensive annotation of 112 functional ncPEPs and specific details regarding the ncRNA transcripts that encode these peptides. We believe that FuncPEP will serve as a platform for further deciphering the biologic significance and medical use of ncPEPs.

The Structure-Function Relationship of Angular Estrogens and Estrogen Receptor Alpha to Initiate Estrogen-Induced Apoptosis in Breast Cancer Cells

High-dose synthetic estrogen therapy was the standard treatment of advanced breast cancer for three decades until the discovery of tamoxifen. A range of substituted triphenylethylene synthetic estrogens and diethylstilbestrol were used. It is now known that low doses of estrogens can cause apoptosis in long-term estrogen deprived (LTED) breast cancer cells resistant to antiestrogens. This action of estrogen can explain the reduced breast cancer incidence in postmenopausal women over 60 who are taking conjugated equine estrogens and the beneficial effect of low-dose estrogen treatment of patients with acquired aromatase inhibitor resistance in clinical trials. To decipher the molecular mechanism of estrogens at the estrogen receptor (ER) complex by different types of estrogens-planar [17β-estradiol (E2)] and angular triphenylethylene (TPE) derivatives-we have synthesized a small series of compounds with either no substitutions on the TPE phenyl ring containing the antiestrogenic side chain of endoxifen or a free hydroxyl. In the first week of treatment with E2 the LTED cells undergo apoptosis completely. By contrast, the test TPE derivatives act as antiestrogens with a free para-hydroxyl on the phenyl ring that contains an antiestrogenic side chain in endoxifen. This inhibits early E2-induced apoptosis if a free hydroxyl is present. No substitution at the site occupied by the antiestrogenic side chain of endoxifen results in early apoptosis similar to planar E2 The TPE compounds recruit coregulators to the ER differentially and predictably, leading to delayed apoptosis in these cells. SIGNIFICANCE STATEMENT: In this paper we investigate the role of the structure-function relationship of a panel of synthetic triphenylethylene (TPE) derivatives and a novel mechanism of estrogen-induced cell death in breast cancer, which is now clinically relevant. Our study indicates that these TPE derivatives, depending on the positioning of the hydroxyl groups, induce various conformations of the estrogen receptor's ligand-binding domain, which in turn produces differential recruitment of coregulators and subsequently different apoptotic effects on the antiestrogen-resistant breast cancer cells.

The MD Anderson Prostate Cancer Patient-derived Xenograft Series (MDA PCa PDX) Captures the Molecular Landscape of Prostate Cancer and Facilitates Marker-driven Therapy Development

PURPOSE

Advances in prostate cancer lag behind other tumor types partly due to the paucity of models reflecting key milestones in prostate cancer progression. Therefore, we develop clinically relevant prostate cancer models.

EXPERIMENTAL DESIGN

Since 1996, we have generated clinically annotated patient-derived xenografts (PDXs; the MDA PCa PDX series) linked to specific phenotypes reflecting all aspects of clinical prostate cancer.

RESULTS

We studied two cell line-derived xenografts and the first 80 PDXs derived from 47 human prostate cancer donors. Of these, 47 PDXs derived from 22 donors are working models and can be expanded either as cell lines (MDA PCa 2a and 2b) or PDXs. The histopathologic, genomic, and molecular characteristics (androgen receptor, ERG, and PTEN loss) maintain fidelity with the human tumor and correlate with published findings. PDX growth response to mouse castration and targeted therapy illustrate their clinical utility. Comparative genomic hybridization and sequencing show significant differences in oncogenic pathways in pairs of PDXs derived from different areas of the same tumor. We also identified a recurrent focal deletion in an area that includes the speckle-type POZ protein-like (SPOPL) gene in PDXs derived from seven human donors of 28 studied (25%). SPOPL is a SPOP paralog, and SPOP mutations define a molecular subclass of prostate cancer. SPOPL deletions are found in 7% of The Cancer Genome Atlas prostate cancers, which suggests that our cohort is a reliable platform for targeted drug development.

CONCLUSIONS

The MDA PCa PDX series is a dynamic resource that captures the molecular landscape of prostate cancers progressing under novel treatments and enables optimization of prostate cancer-specific, marker-driven therapy.

Interactive Clustered Heat Map Builder: An easy web-based tool for creating sophisticated clustered heat maps

Clustered heat maps are the most frequently used graphics for visualization and interpretation of genome-scale molecular profiling data in biology.  Construction of a heat map generally requires the assistance of a biostatistician or bioinformatics analyst capable of working in R or a similar programming language to transform the study data, perform hierarchical clustering, and generate the heat map.  Our web-based Interactive Heat Map Builder can be used by investigators with no bioinformatics experience to generate high-caliber, publication quality maps.  Preparation of the data and construction of a heat map is rarely a simple linear process.  Our tool allows a user to move back and forth iteratively through the various stages of map generation to try different options and approaches.  Finally, the heat map the builder creates is available in several forms, including an interactive Next-Generation Clustered Heat Map that can be explored dynamically to investigate the results more fully.

Interactive Clustered Heat Map Builder: An easy web-based tool for creating sophisticated clustered heat maps

Clustered heat maps are the most frequently used graphics for visualization and interpretation of genome-scale molecular profiling data in biology.  Construction of a heat map generally requires the assistance of a biostatistician or bioinformatics analyst capable of working in R or a similar programming language to transform the study data, perform hierarchical clustering, and generate the heat map.  Our web-based Interactive Heat Map Builder can be used by investigators with no bioinformatics experience to generate high-caliber, publication quality maps.  Preparation of the data and construction of a heat map is rarely a simple linear process.  Our tool allows a user to move back and forth iteratively through the various stages of map generation to try different options and approaches.  Finally, the heat map the builder creates is available in several forms, including an interactive Next-Generation Clustered Heat Map that can be explored dynamically to investigate the results more fully.

Back to the Colorectal Cancer Consensus Molecular Subtype Future

PURPOSE OF REVIEW

This review seeks to provide an informed prospective on the advances in molecular profiling and analysis of colorectal cancer (CRC). The goal is to provide a historical context and current summary on how advances in gene and protein sequencing technology along with computer capabilities led to our current bioinformatic advances in the field.

RECENT FINDINGS

An explosion of knowledge has occurred regarding genetic, epigenetic, and biochemical alterations associated with the evolution of colorectal cancer. This has led to the realization that CRC is a heterogeneous disease with molecular alterations often dictating natural history, response to treatment, and outcome. The consensus molecular subtypes (CMS) classification classifies CRC into four molecular subtypes with distinct biological characteristics, which may form the basis for clinical stratification and subtype-based targeted intervention. This review summarizes new developments of a field moving "Back to the Future." CRC molecular subtyping will better identify key subtype specific therapeutic targets and responses to therapy.

Targeting the MYCN-PARP-DNA Damage Response Pathway in Neuroendocrine Prostate Cancer

Purpose: We investigated MYCN-regulated molecular pathways in castration-resistant prostate cancer (CRPC) classified by morphologic criteria as adenocarcinoma or neuroendocrine to extend the molecular phenotype, establish driver pathways, and identify novel approaches to combination therapy for neuroendocrine prostate cancer (NEPC).Experimental Design and Results: Using comparative bioinformatics analyses of CRPC-Adeno and CRPC-Neuro RNA sequence data from public data sets and a panel of 28 PDX models, we identified a MYCN-PARP-DNA damage response (DDR) pathway that is enriched in CRPC with neuroendocrine differentiation (NED) and CRPC-Neuro. ChIP-PCR assay revealed that N-MYC transcriptionally activates PARP1, PARP2, BRCA1, RMI2, and TOPBP1 through binding to the promoters of these genes. MYCN or PARP1 gene knockdown significantly reduced the expression of MYCN-PARP-DDR pathway genes and NED markers, and inhibition with MYCNsi and/or PARPsi, BRCA1si, or RMI2si significantly suppressed malignant activities, including cell viability, colony formation, and cell migration, in C4-2b4 and NCI-H660 cells. Targeting this pathway with AURKA inhibitor PHA739358 and PARP inhibitor olaparib generated therapeutic effects similar to those of gene knockdown in vitro and significantly suppressed tumor growth in both C4-2b4 and MDACC PDX144-13C subcutaneous models in vivoConclusions: Our results identify a novel MYCN-PARP-DDR pathway that is driven by N-MYC in a subset of CRPC-Adeno and in NEPC. Targeting this pathway using in vitro and in vivo CRPC-Adeno and CRPC-Neuro models demonstrated a novel therapeutic strategy for NEPC. Further investigation of N-MYC-regulated DDR gene targets and the biological and clinical significance of MYCN-PARP-DDR signaling will more fully elucidate the importance of the MYCN-PARP-DDR signaling pathway in the development and maintenance of NEPC. Clin Cancer Res; 24(3); 696-707. ©2017 AACR.

A Galaxy Implementation of Next-Generation Clustered Heatmaps for Interactive Exploration of Molecular Profiling Data

Clustered heatmaps are the most frequently used graphics for visualization of molecular profiling data in biology. However, they are generally rendered as static, or only modestly interactive, images. We have now used recent advances in web technologies to produce interactive "next-generation" clustered heatmaps (NG-CHM) that enable extreme zooming and navigation without loss of resolution. NG-CHMs also provide link-outs to additional information sources and include other features that facilitate deep exploration of the biology behind the image. Here, we describe an implementation of the NG-CHM system in the Galaxy bioinformatics platform. We illustrate the algorithm and available computational tool using RNA-seq data from The Cancer Genome Atlas program's Kidney Clear Cell Carcinoma project. Cancer Res; 77(21); e23-26. ©2017 AACR.

NSD1 Inactivation and SETD2 Mutation Drive a Convergence toward Loss of Function of H3K36 Writers in Clear Cell Renal Cell Carcinomas

Extensive dysregulation of chromatin-modifying genes in clear cell renal cell carcinoma (ccRCC) has been uncovered through next-generation sequencing. However, a scientific understanding of the cross-talk between epigenetic and genomic aberrations remains limited. Here we identify three ccRCC epigenetic clusters, including a clear cell CpG island methylator phenotype (C-CIMP) subgroup associated with promoter methylation of VEGF genes (FLT4, FLT1, and KDR). C-CIMP was furthermore characterized by silencing of genes related to vasculature development. Through an integrative analysis, we discovered frequent silencing of the histone H3 K36 methyltransferase NSD1 as the sole chromatin-modifying gene silenced by DNA methylation in ccRCC. Notably, tumors harboring NSD1 methylation were of higher grade and stage in different ccRCC datasets. NSD1 promoter methylation correlated with SETD2 somatic mutations across and within spatially distinct regions of primary ccRCC tumors. ccRCC harboring epigenetic silencing of NSD1 displayed a specific genome-wide methylome signature consistent with the NSD1 mutation methylome signature observed in Sotos syndrome. Thus, we concluded that epigenetic silencing of genes involved in angiogenesis is a hallmark of the methylator phenotype in ccRCC, implying a convergence toward loss of function of epigenetic writers of the H3K36 histone mark as a root feature of aggressive ccRCC. Cancer Res; 77(18); 4835-45. ©2017 AACR.

Androgen receptor inhibitor-induced "BRCAness" and PARP inhibition are synthetically lethal for castration-resistant prostate cancer

Cancers with loss-of-function mutations in BRCA1 or BRCA2 are deficient in the DNA damage repair pathway called homologous recombination (HR), rendering these cancers exquisitely vulnerable to poly(ADP-ribose) polymerase (PARP) inhibitors. This functional state and therapeutic sensitivity is referred to as "BRCAness" and is most commonly associated with some breast cancer types. Pharmaceutical induction of BRCAness could expand the use of PARP inhibitors to other tumor types. For example, BRCA mutations are present in only ~20% of prostate cancer patients. We found that castration-resistant prostate cancer (CRPC) cells showed increased expression of a set of HR-associated genes, including BRCA1, RAD54L, and RMI2 Although androgen-targeted therapy is typically not effective in CRPC patients, the androgen receptor inhibitor enzalutamide suppressed the expression of those HR genes in CRPC cells, thus creating HR deficiency and BRCAness. A "lead-in" treatment strategy, in which enzalutamide was followed by the PARP inhibitor olaparib, promoted DNA damage-induced cell death and inhibited clonal proliferation of prostate cancer cells in culture and suppressed the growth of prostate cancer xenografts in mice. Thus, antiandrogen and PARP inhibitor combination therapy may be effective for CRPC patients and suggests that pharmaceutically inducing BRCAness may expand the clinical use of PARP inhibitors.

Modeling of Patient-Derived Xenografts in Colorectal Cancer

Developing realistic preclinical models using clinical samples that mirror complex tumor biology and behavior are vital to advancing cancer research. While cell line cultures have been helpful in generating preclinical data, the genetic divergence between these and corresponding primary tumors has limited clinical translation. Conversely, patient-derived xenografts (PDX) in colorectal cancer are highly representative of the genetic and phenotypic heterogeneity in the original tumor. Coupled with high-throughput analyses and bioinformatics, these PDXs represent robust preclinical tools for biomarkers, therapeutic target, and drug discovery. Successful PDX engraftment is hypothesized to be related to a series of anecdotal variables namely, tissue source, cancer stage, tumor grade, acquisition strategy, time to implantation, exposure to prior systemic therapy, and genomic heterogeneity of tumors. Although these factors at large can influence practices and patterns related to xenotransplantation, their relative significance in determining the success of establishing PDXs is uncertain. Accordingly, we systematically examined the predictive ability of these factors in establishing PDXs using 90 colorectal cancer patient specimens that were subcutaneously implanted into immunodeficient mice. Fifty (56%) PDXs were successfully established. Multivariate analyses showed tissue acquisition strategy [surgery 72.0% (95% confidence interval (CI): 58.2-82.6) vs. biopsy 35% (95% CI: 22.1%-50.6%)] to be the key determinant for successful PDX engraftment. These findings contrast with current empiricism in generating PDXs and can serve to simplify or liberalize PDX modeling protocols. Better understanding the relative impact of these factors on efficiency of PDX formation will allow for pervasive integration of these models in care of colorectal cancer patients. Mol Cancer Ther; 16(7); 1435-42. ©2017 AACR.

PathwaysWeb: a gene pathways API with directional interactions, expanded gene ontology, and versioning

PathwaysWeb is a resource-based, well-documented web system that provides publicly available information on genes, biological pathways, Gene Ontology (GO) terms, gene-gene interaction networks (importantly, with the directionality of interactions) and links to key-related PubMed documents. The PathwaysWeb API simplifies the construction of applications that need to retrieve and interrelate information across multiple, pathway-related data types from a variety of original data sources. PathwaysBrowser is a companion website that enables users to explore the same integrated pathway data. The PathwaysWeb system facilitates reproducible analyses by providing access to all versions of the integrated datasets. Although its GO subsystem includes data for mouse, PathwaysWeb currently focuses on human data. However, pathways for mouse and many other species can be inferred with a high success rate from human pathways.

AVAILABILITY AND IMPLEMENTATION

PathwaysWeb can be accessed via the Internet at http://bioinformatics.mdanderson.org/main/PathwaysWeb:Overview.

CONTACT

jmmelott@mdanderson.org

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Abstract 2979: A web portal of ‘next-generation’ clustered heat maps for user-friendly, interactive exploration of patterns in TCGA data

The Cancer Genome Atlas (TCGA) program is generating comprehensive molecular profiles of more than 30 clinical tumor types, the first 12 of which have been incorporated into a “Pan-Cancer12” project. One bioinformatic challenge is statistical analysis of the resulting profiles; a second is the visual detective work necessary to explore individual genes, pathways and patterns in the data. For that type of detective work, we introduced CHMs in the early 1990s for pharmacogenomic analysis (1) and later for integrated visualization of genomic, transcriptomic, proteomic, pharmacological, and functional data (2). As the ubiquitous first-order way of visualizing omic data, CHMs have appeared in many thousands of publications (3-9), including all of the major publications by the TCGA Research Network. However, a major limitation is that they have been static or only modestly interactive graphics. We have now developed “next-generation” clustered heat maps (NG-CHMs), which use a Google-maps-like tiling technology for extreme zooming and navigation without loss of resolution. NG-CHMs provide pathway and gene ontology information, re-coloring on the fly, tools for reproducibility, high-resolution graphics output, a statistical toolbox, and link-outs to public sources of information on genes, proteins, pathways and drugs. The result is a visually rich, dynamic environment for exploration of the masses of data produced by TCGA. The compendium of TCGA Pan-Cancer NG-CHMs currently includes 667 maps as an initial set, but the numbers will soon rise into the thousands as more data types, tumour types and algorithms are incorporated (at web portal http://bioinformatics.mdanderson.org/TCGA/NGCHMPortal/). As an illustrative example, NG-CHMs proved pivotal as a tool for discovering and analyzing molecular target themes common to multiple types of gynecological cancers and themes that distinguish them from each other.

1. Weinstein JN … Paull KD. Stem Cells 12; 13, 1994.

2. Weinstein JN … Paull KD. Science 275;343, 1997.

3. Myers TG … Weinstein JN. Electrophoresis 18; 467, 1997.

4. Eisen MB … Botstein D. Proc. Natl. Acad. Sci. U.S.A. 14863, 1998.

5. Golub TR … Lander ES. Science 286; 531, 1999.

6. Ross DT … Brown PA. Nature Genetics 24; 227, 2000

7. Scherf U … Weinstein JN. Nature Genetics 24; 236, 2000.

8. Zeeberg BR … Weinstein JN. BMC Bioinformatics 6; 168, 2005.

9. Weinstein JN. Science 319; 1772, 2008.

Supported in part by NCI Grant No. U24CA143883, by a gift from the Mary K. Chapman Foundation, and by a grant from the Michael and Susan Dell Foundation honoring Lorraine Dell.

Note: This abstract was not presented at the meeting.

Citation Format: John N. Weinstein, Rehan Akbani, David W. Kane, James M. Melott, Tod D. Casasent, Rong Yao, Paul L. Roebuck, Gordon B. Mills, Michael C. Ryan, Christopher Wakefield, Bradley M. Broom. A web portal of ‘next-generation’ clustered heat maps for user-friendly, interactive exploration of patterns in TCGA data. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2979. doi:10.1158/1538-7445.AM2015-2979

Inhibition of mTORC1/2 overcomes resistance to MAPK pathway inhibitors mediated by PGC1α and oxidative phosphorylation in melanoma

Metabolic heterogeneity is a key factor in cancer pathogenesis. We found that a subset of BRAF- and NRAS-mutant human melanomas resistant to the MEK inhibitor selumetinib displayed increased oxidative phosphorylation (OxPhos) mediated by the transcriptional coactivator PGC1α. Notably, all selumetinib-resistant cells with elevated OxPhos could be resensitized by cotreatment with the mTORC1/2 inhibitor AZD8055, whereas this combination was ineffective in resistant cell lines with low OxPhos. In both BRAF- and NRAS-mutant melanoma cells, MEK inhibition increased MITF expression, which in turn elevated levels of PGC1α. In contrast, mTORC1/2 inhibition triggered cytoplasmic localization of MITF, decreasing PGC1α expression and inhibiting OxPhos. Analysis of tumor biopsies from patients with BRAF-mutant melanoma progressing on BRAF inhibitor ± MEK inhibitor revealed that PGC1α levels were elevated in approximately half of the resistant tumors. Overall, our findings highlight the significance of OxPhos in melanoma and suggest that combined targeting of the MAPK and mTORC pathways may offer an effective therapeutic strategy to treat melanomas with this metabolic phenotype.

Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia

BACKGROUND

Many mutations that contribute to the pathogenesis of acute myeloid leukemia (AML) are undefined. The relationships between patterns of mutations and epigenetic phenotypes are not yet clear.

METHODS

We analyzed the genomes of 200 clinically annotated adult cases of de novo AML, using either whole-genome sequencing (50 cases) or whole-exome sequencing (150 cases), along with RNA and microRNA sequencing and DNA-methylation analysis.

RESULTS

AML genomes have fewer mutations than most other adult cancers, with an average of only 13 mutations found in genes. Of these, an average of 5 are in genes that are recurrently mutated in AML. A total of 23 genes were significantly mutated, and another 237 were mutated in two or more samples. Nearly all samples had at least 1 nonsynonymous mutation in one of nine categories of genes that are almost certainly relevant for pathogenesis, including transcription-factor fusions (18% of cases), the gene encoding nucleophosmin (NPM1) (27%), tumor-suppressor genes (16%), DNA-methylation-related genes (44%), signaling genes (59%), chromatin-modifying genes (30%), myeloid transcription-factor genes (22%), cohesin-complex genes (13%), and spliceosome-complex genes (14%). Patterns of cooperation and mutual exclusivity suggested strong biologic relationships among several of the genes and categories.

CONCLUSIONS

We identified at least one potential driver mutation in nearly all AML samples and found that a complex interplay of genetic events contributes to AML pathogenesis in individual patients. The databases from this study are widely available to serve as a foundation for further investigations of AML pathogenesis, classification, and risk stratification. (Funded by the National Institutes of Health.).

Abstract 5132: Interactively exploring patterns in TCGA data: a web-based compendium of ‘next-generation’ clustered heat maps

Each of the 5 TCGA marker paper published in Nature to date has included at least one clustered heat map (CHM). We introduced CHMs in the early 1990’s for pharmacogenomic analysis (1) and later for integrated visualization of genomic, transcriptomic, proteomic, pharmacological, and functional data (1). As the ubiquitous first-order way of visualizing omic data, CHMs have appeared in many thousands of publications (3–9), including those from TCGA. We have elsewhere summarized their limitations (10).

One such limitation is that CHMs are generally static images. We therefore initiated the next-generation CHM (NG-CHM) project, using an image-tiling technology similar to that in Google Maps for navigation and extreme drill-down without loss of resolution. Once the CHM has been zoomed sufficiently, labels (e.g., gene, protein, or drug names) appear on the image's axes. Clicking on a label produces a menu of link-outs (e.g., to GeneCards, Google, PubMed). For gene vs. gene maps, each pixel can represent a color-coded Pearson correlation coefficient. Clicking on the pixel pulls up the corresponding data scattergram, bootstrap statistics, literature references, or pathway relationships. Strong usability features include floating windows, flexible search tools, cluster selection tools, customizable re-coloring of the CHM, and high-quality PDF's suitable for publication. NG-CHMs are a major resource for exploratory analysis and visualization in multiple projects of TCGA and other large-scale molecular profiling programs. Explore interactive versions for TCGA breast, colorectal, lung squamous, and glioblastoma data at http://bioinformatics.mdanderson.org/main/TCGA/NGCHM.

Supported in part by NCI Grant No. U24CA143883, by a gift from the Mary K. Chapman Foundation, and by a grant from the Michael and Susan Dell Foundation honoring Lorraine Dell.

Citation Format: John N. Weinstein, David W. Kane, Rehan Akbani, Deepti Dodda, Lam Nguyen, Michael C. Ryan, Chris Wakefield, Bradley M. Broom. Interactively exploring patterns in TCGA data: a web-based compendium of ‘next-generation’ clustered heat maps. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 5132. doi:10.1158/1538-7445.AM2013-5132

iBAG: integrative Bayesian analysis of high-dimensional multiplatform genomics data

Motivation: Analyzing data from multi-platform genomics experiments combined with patients’ clinical outcomes helps us understand the complex biological processes that characterize a disease, as well as how these processes relate to the development of the disease. Current data integration approaches are limited in that they do not consider the fundamental biological relationships that exist among the data obtained from different platforms.

Statistical Model: We propose an integrative Bayesian analysis of genomics data (iBAG) framework for identifying important genes/biomarkers that are associated with clinical outcome. This framework uses hierarchical modeling to combine the data obtained from multiple platforms into one model.

Results: We assess the performance of our methods using several synthetic and real examples. Simulations show our integrative methods to have higher power to detect disease-related genes than non-integrative methods. Using the Cancer Genome Atlas glioblastoma dataset, we apply the iBAG model to integrate gene expression and methylation data to study their associations with patient survival. Our proposed method discovers multiple methylation-regulated genes that are related to patient survival, most of which have important biological functions in other diseases but have not been previously studied in glioblastoma.

Availability:http://odin.mdacc.tmc.edu/∼vbaladan/.

Contact:veera@mdanderson.org

Supplementary information:Supplementary data are available at Bioinformatics online.

Model averaging strategies for structure learning in Bayesian networks with limited data

BACKGROUND

Considerable progress has been made on algorithms for learning the structure of Bayesian networks from data. Model averaging by using bootstrap replicates with feature selection by thresholding is a widely used solution for learning features with high confidence. Yet, in the context of limited data many questions remain unanswered. What scoring functions are most effective for model averaging? Does the bias arising from the discreteness of the bootstrap significantly affect learning performance? Is it better to pick the single best network or to average multiple networks learnt from each bootstrap resample? How should thresholds for learning statistically significant features be selected?

RESULTS

The best scoring functions are Dirichlet Prior Scoring Metric with small λ and the Bayesian Dirichlet metric. Correcting the bias arising from the discreteness of the bootstrap worsens learning performance. It is better to pick the single best network learnt from each bootstrap resample. We describe a permutation based method for determining significance thresholds for feature selection in bagged models. We show that in contexts with limited data, Bayesian bagging using the Dirichlet Prior Scoring Metric (DPSM) is the most effective learning strategy, and that modifying the scoring function to penalize complex networks hampers model averaging. We establish these results using a systematic study of two well-known benchmarks, specifically ALARM and INSURANCE. We also apply our network construction method to gene expression data from the Cancer Genome Atlas Glioblastoma multiforme dataset and show that survival is related to clinical covariates age and gender and clusters for interferon induced genes and growth inhibition genes.

CONCLUSIONS

For small data sets, our approach performs significantly better than previously published methods.

Selective genomic copy number imbalances and probability of recurrence in early-stage breast cancer

A number of studies of copy number imbalances (CNIs) in breast tumors support associations between individual CNIs and patient outcomes. However, no pattern or signature of CNIs has emerged for clinical use. We determined copy number (CN) gains and losses using high-density molecular inversion probe (MIP) arrays for 971 stage I/II breast tumors and applied a boosting strategy to fit hazards models for CN and recurrence, treating chromosomal segments in a dose-specific fashion (-1 [loss], 0 [no change] and +1 [gain]). The concordance index (C-Index) was used to compare prognostic accuracy between a training (n = 728) and test (n = 243) set and across models. Twelve novel prognostic CNIs were identified: losses at 1p12, 12q13.13, 13q12.3, 22q11, and Xp21, and gains at 2p11.1, 3q13.12, 10p11.21, 10q23.1, 11p15, 14q13.2-q13.3, and 17q21.33. In addition, seven CNIs previously implicated as prognostic markers were selected: losses at 8p22 and 16p11.2 and gains at 10p13, 11q13.5, 12p13, 20q13, and Xq28. For all breast cancers combined, the final full model including 19 CNIs, clinical covariates, and tumor marker-approximated subtypes (estrogen receptor [ER], progesterone receptor, ERBB2 amplification, and Ki67) significantly outperformed a model containing only clinical covariates and tumor subtypes (C-Index(full model), train[test]  =  0.72[0.71] ± 0.02 vs. C-Index(clinical + subtype model), train[test]  =  0.62[0.62] ± 0.02; p<10(-6)). In addition, the full model containing 19 CNIs significantly improved prognostication separately for ER-, HER2+, luminal B, and triple negative tumors over clinical variables alone. In summary, we show that a set of 19 CNIs discriminates risk of recurrence among early-stage breast tumors, independent of ER status. Further, our data suggest the presence of specific CNIs that promote and, in some cases, limit tumor spread.

Bagged gene shaving for the robust clustering of high-throughput data

The analysis of high-throughput data sets, such as microarray data, often requires that individual variables (genes, for example) be grouped into clusters of variables with highly correlated values across all samples. Gene shaving is an established method for generating such clusters, but is overly sensitive to the input data: changing just one sample can determine whether or not an entire cluster is found. This paper describes a clustering method based on the bootstrap aggregation of gene shaving clusters, which overcomes this and other problems, and applies the new method to a large gene expression microarray dataset from brain tumour samples.

Bayesian ensemble methods for survival prediction in gene expression data

MOTIVATION

We propose a Bayesian ensemble method for survival prediction in high-dimensional gene expression data. We specify a fully Bayesian hierarchical approach based on an ensemble 'sum-of-trees' model and illustrate our method using three popular survival models. Our non-parametric method incorporates both additive and interaction effects between genes, which results in high predictive accuracy compared with other methods. In addition, our method provides model-free variable selection of important prognostic markers based on controlling the false discovery rates; thus providing a unified procedure to select relevant genes and predict survivor functions.

RESULTS

We assess the performance of our method several simulated and real microarray datasets. We show that our method selects genes potentially related to the development of the disease as well as yields predictive performance that is very competitive to many other existing methods.

AVAILABILITY

http://works.bepress.com/veera/1/.

Building networks with microarray data

This chapter describes methods for learning gene interaction networks from high-throughput gene expression data sets. Many genes have unknown or poorly understood functions and interactions, especially in diseases such as cancer where the genome is frequently mutated. The gene interactions inferred by learning a network model from the data can form the basis of hypotheses that can be verified by subsequent biological experiments. This chapter focuses specifically on Bayesian network models, which have a level of mathematical detail greater than purely conceptual models but less than detailed differential equation models. From a network learning perspective the most severe problem with microarray data is the limited sample size, since there are usually many plausible networks for modeling the system. Since these cannot be reliably distinguished using the number of samples found in current microarray data sets, we describe robust network learning strategies for reducing the number of false interactions detected. We perform preliminary clustering using co-expression network analysis and gene shaving. Subsequently we construct Bayesian networks to obtain a global perspective of the relationships between these gene clusters. Throughout this chapter, we illustrate the concepts being expounded by referring to an ongoing example of a publicly available breast cancer data set.

Gene expression meta-analysis supports existence of molecular apocrine breast cancer with a role for androgen receptor and implies interactions with ErbB family

Background

Pathway discovery from gene expression data can provide important insight into the relationship between signaling networks and cancer biology. Oncogenic signaling pathways are commonly inferred by comparison with signatures derived from cell lines. We use the Molecular Apocrine subtype of breast cancer to demonstrate our ability to infer pathways directly from patients' gene expression data with pattern analysis algorithms.

Methods

We combine data from two studies that propose the existence of the Molecular Apocrine phenotype. We use quantile normalization and XPN to minimize institutional bias in the data. We use hierarchical clustering, principal components analysis, and comparison of gene signatures derived from Significance Analysis of Microarrays to establish the existence of the Molecular Apocrine subtype and the equivalence of its molecular phenotype across both institutions. Statistical significance was computed using the Fasano & Franceschini test for separation of principal components and the hypergeometric probability formula for significance of overlap in gene signatures. We perform pathway analysis using LeFEminer and Backward Chaining Rule Induction to identify a signaling network that differentiates the subset. We identify a larger cohort of samples in the public domain, and use Gene Shaving and Robust Bayesian Network Analysis to detect pathways that interact with the defining signal.

Results

We demonstrate that the two separately introduced ER^- breast cancer subsets represent the same tumor type, called Molecular Apocrine breast cancer. LeFEminer and Backward Chaining Rule Induction support a role for AR signaling as a pathway that differentiates this subset from others. Gene Shaving and Robust Bayesian Network Analysis detect interactions between the AR pathway, EGFR trafficking signals, and ErbB2.

Conclusion

We propose criteria for meta-analysis that are able to demonstrate statistical significance in establishing molecular equivalence of subsets across institutions. Data mining strategies used here provide an alternative method to comparison with cell lines for discovering seminal pathways and interactions between signaling networks. Analysis of Molecular Apocrine breast cancer implies that therapies targeting AR might be hampered if interactions with ErbB family members are not addressed.

Learning robust cell signalling models from high throughput proteomic data

We propose a framework for learning robust Bayesian network models of cell signalling from high-throughput proteomic data. We show that model averaging using Bayesian bootstrap resampling generates more robust structures than procedures that learn structures using all of the data. We also develop an algorithm for ranking the importance of network features using bootstrap resample data. We apply our algorithms to derive the T-cell signalling network from the flow cytometry data of Sachs et al. (2005). Our learning algorithm has identified, with high confidence, several new crosstalk mechanisms in the T-cell signalling network. Many of them have already been confirmed experimentally in the recent literature and six new crosstalk mechanisms await experimental validation.

Predicting altered pathways using extendable scaffolds

Many diseases, especially solid tumors, involve the disruption or deregulation of cellular processes. Most current work using gene expression and other high-throughput data, simply list a set of differentially expressed genes. We propose a new method, PAPES (predicting altered pathways using extendable scaffolds), to computationally reverse-engineer models of biological systems. We use sets of genes that occur in a known biological pathway to construct component process models. We then compose these models to build larger scale networks that capture interactions among pathways. We show that we can learn process modifications in two coupled metabolic pathways in prostate cancer cells.

Genetic analysis of the age at menopause by using estimating equations and Bayesian random effects models

Multi-wave self-report data on age at menopause in 2182 female twin pairs (1355 monozygotic and 827 dizygotic pairs), were analysed to estimate the genetic, common and unique environmental contribution to variation in age at menopause. Two complementary approaches for analysing correlated time-to-onset twin data are considered: the generalized estimating equations (GEE) method in which one can estimate zygosity-specific dependence simultaneously with regression coefficients that describe the average population response to changing covariates; and a subject-specific Bayesian mixed model in which heterogeneity in regression parameters is explicitly modelled and the different components of variation may be estimated directly. The proportional hazards and Weibull models were utilized, as both produce natural frameworks for estimating relative risks while adjusting for simultaneous effects of other covariates. A simple Markov chain Monte Carlo method for covariate imputation of missing data was used and the actual implementation of the Bayesian model was based on Gibbs sampling using the freeware package BUGS.