The Cancer Genome Atlas (TCGA) research network has made public a large collection of clinical and molecular phenotypes of more than 10 000 tumor patients across 33 different tumor types. Using this cohort, TCGA has published over 20 marker papers detailing the genomic and epigenomic alterations associated with these tumor types. Although many important discoveries have been made by TCGA's research network, opportunities still exist to implement novel methods, thereby elucidating new biological pathways and diagnostic markers. However, mining the TCGA data presents several bioinformatics challenges, such as data retrieval and integration with clinical data and other molecular data types (e.g. RNA and DNA methylation). We developed an R/Bioconductor package called TCGAbiolinks to address these challenges and offer bioinformatics solutions by using a guided workflow to allow users to query, download and perform integrative analyses of TCGA data. We combined methods from computer science and statistics into the pipeline and incorporated methodologies developed in previous TCGA marker studies and in our own group. Using four different TCGA tumor types (Kidney, Brain, Breast and Colon) as examples, we provide case studies to illustrate examples of reproducibility, integrative analysis and utilization of different Bioconductor packages to advance and accelerate novel discoveries.

Species interaction networks are shaped by abiotic and biotic factors. Here, as part of theTaraOceans project, we studied the photic zone interactome using environmental factors and organismal abundance profiles and found that environmental factors are incomplete predictors of community structure. We found associations across plankton functional types and phylogenetic groups to be nonrandomly distributed on the network and driven by both local and global patterns. We identified interactions among grazers, primary producers, viruses, and (mainly parasitic) symbionts and validated network-generated hypotheses using microscopy to confirm symbiotic relationships. We have thus provided a resource to support further research on ocean food webs and integrating biological components into ocean models.

Purpose: Recently, several prognostic gene expression signatures have been identified; however, their performance has never been evaluated according to the previously described molecular subtypes based on the estrogen receptor (ER) and human epidermal growth factor receptor 2 (HER2), and their biological meaning has remained unclear. Here we aimed to perform a comprehensive meta-analysis integrating both clinicopathologic and gene expression data, focusing on the main molecular subtypes. Experimental Design: We developed gene expression modules related to key biological processes in breast cancer such as tumor invasion, immune response, angiogenesis, apoptosis, proliferation, and ER and HER2 signaling, and then analyzed these modules together with clinical variables and several prognostic signatures on publicly available microarray studies (>2,100 patients). Results: Multivariate analysis showed that in the ER+/HER2− subgroup, only the proliferation module and the histologic grade were significantly associated with clinical outcome. In the ER−/HER2− subgroup, only the immune response module was associated with prognosis, whereas in the HER2+ tumors, the tumor invasion and immune response modules displayed significant association with survival. Proliferation was identified as the most important component of several prognostic signatures, and their performance was limited to the ER+/HER2− subgroup. Conclusions: Although proliferation is the strongest parameter predicting clinical outcome in the ER+/HER2− subtype and the common denominator of most prognostic gene signatures, immune response and tumor invasion seem to be the main molecular processes associated with prognosis in the ER−/HER2− and HER2+ subgroups, respectively. These findings may help to define new clinicogenomic models and to identify new therapeutic strategies in the specific molecular subgroups.

Purpose: A number of microarray studies have reported distinct molecular profiles of breast cancers (BC), such as basal-like, ErbB2-like, and two to three luminal-like subtypes. These were associated with different clinical outcomes. However, although the basal and the ErbB2 subtypes are repeatedly recognized, identification of estrogen receptor (ER) –positive subtypes has been inconsistent. Therefore, refinement of their molecular definition is needed. Materials and Methods: We have previously reported a gene expression grade index (GGI), which defines histologic grade based on gene expression profiles. Using this algorithm, we assigned ER-positive BC to either high–or low–genomic grade subgroups and compared these with previously reported ER-positive molecular classifications. As further validation, we classified 666 ER-positive samples into subtypes and assessed their clinical outcome. Results: Two ER-positive molecular subgroups (high and low genomic grade) could be defined using the GGI. Despite tracking a single biologic pathway, these were highly comparable to the previously described luminal A and B classification and significantly correlated to the risk groups produced using the 21-gene recurrence score. The two subtypes were associated with statistically distinct clinical outcome in both systemically untreated and tamoxifen-treated populations. Conclusion: The use of genomic grade can identify two clinically distinct ER-positive molecular subtypes in a simple and highly reproducible manner across multiple data sets. This study emphasizes the important role of proliferation-related genes in predicting prognosis in ER-positive BC.

Detecting frauds in credit card transactions is perhaps one of the best testbeds for computational intelligence algorithms. In fact, this problem involves a number of relevant challenges, namely: concept drift (customers' habits evolve and fraudsters change their strategies over time), class imbalance (genuine transactions far outnumber frauds), and verification latency (only a small set of transactions are timely checked by investigators). However, the vast majority of learning algorithms that have been proposed for fraud detection rely on assumptions that hardly hold in a real-world fraud-detection system (FDS). This lack of realism concerns two main aspects: 1) the way and timing with which supervised information is provided and 2) the measures used to assess fraud-detection performance. This paper has three major contributions. First, we propose, with the help of our industrial partner, a formalization of the fraud-detection problem that realistically describes the operating conditions of FDSs that everyday analyze massive streams of credit card transactions. We also illustrate the most appropriate performance measures to be used for fraud-detection purposes. Second, we design and assess a novel learning strategy that effectively addresses class imbalance, concept drift, and verification latency. Third, in our experiments, we demonstrate the impact of class unbalance and concept drift in a real-world data stream containing more than 75 million transactions, authorized over a time window of three years.