oposSOM-Browser: an interactive tool to explore omics data landscapes in health science

Background oposSOM is a comprehensive, machine learning based open-source data analysis software combining functionalities such as diversity analyses, biomarker selection, function mining, and visualization. Results These functionalities are now available as interactive web-browser application for a broader user audience interested in extracting detailed information from high-throughput omics data sets pre-processed by oposSOM. It enables interactive browsing of single-gene and gene set profiles, of molecular ‘portrait landscapes’, of associated phenotype diversity, and signalling pathway activation patterns. Conclusion The oposSOM-Browser makes available interactive data browsing for five transcriptome data sets of cancer (melanomas, B-cell lymphomas, gliomas) and of peripheral blood (sepsis and healthy individuals) at www.izbi.uni-leipzig.de/opossom-browser.

The different data sets provided by oposSOM-Browsers are accessed via a select box in the top panel, which also includes version and citation information, and the main navigation menu ( Figure S 1). The overview tab gives general information about the data set currently selected, such as dimensionality and version of oposSOM package used for data processing.

Map browser
The map browser provides information about expression modules detected in the data landscape, and of mapping of age, gender, and survival information into the data landscape.
Module and data maps can be selected on the left-hand side ( Figure S 4).
In the module maps, individual expression modules can be selected in the center frame, which will generate the corresponding expression profile across all samples of the study (Figure S 4).
Additionally, the table on the right-hand side listsdepending on the selected tab -all genes in the module, enriched functional gene sets, or the subgroups which show high expression of this module, respectively.
In the data maps, single metagenes (i.e. pixels in the mosaic map) can be selected (Figure S 5).
Additional information will be shown for all participants expressing this metagene: In the age map, age distribution of the corresponding participants will be shown in terms of a histogram, in the sex map a pie chart, and in the survival map curves of overall survival of patients expressing this metagene compared to those not expressing the metagene. The right-hand table lists all genes and enriched functional gene sets in the metagene, and the subgroups of the participants expressing this metagene, respectively.

Phenotype browser
The phenotype browser allows for the evaluation of different phenotypic stratifications with regard to the sample landscape and to prognostic impact ( Figure S 6). General, clinical, or molecular phenotypes can be selected in the top left menu, where also individual phenotype characteristics can be added or removed.
The sample landscape, a correlation network of all samples in the data set, is then colored according to the selection. Clicking into one of the network nodes will open a frame showing the sample expression portrait along with patient age and gender if available.
Curves of overall survival as function of time are also generated given this information is available in the data set. They are stratified by the selected phenotype characteristics to evaluate associated diversification of prognosis. In the signature browser, individual lists of genes can be uploaded to benchmark classification using these genes ( Figure S 7). Click the gear button to open the menu for uploading the signature genes. They can be provided as Ensemble-IDs or gene names. For a submitted gene signature, the expression profile across all samples and the mapping into the data landscape will be displayed.
The signature is then used to identify samples of the target class in a set of samples that comprises the target class and additional classes/characteristics as selected in the bottom left select boxes. After selection, a click on the 'Generate ROC' button will start benchmarking resulting in a ROC curve plot together with the corresponding AUC measure.