Accelerative integromics: dynamic biomarker discovery using big data.
Session type: Poster / e-Poster / Silent Theatre session
Theme: Diagnosis and therapy
Recent advancements in high throughput technologies have enabled rapid and affordable profiling of cancer tumours. However, the majority of the subsequent outputs are held in isolated silos, and are analysed as separate entities decreasing information discovery. On contrast, facilitating the integration of multiple ‘-omics’ data (integromics), would create a more holistic and multidimensional view of biological systems and disease phenotypes. To this end our aim was to develop a dynamic platform capable of computational integration and analysis, enabling biomarker discovery through big data analysis.
The platform was provided to the users in the form of a searchable web interface application. To provide best statistical analysis processes, the R statistical framework was used to support the back-end algorithms, coupled with a NoSQL database management system, to maintain efficiency, while minimising data redundancy. A dataset of 264 prostate cancer microarray expression samples, with clinicopathological data (121,563 probes), were used to benchmark the performance gains of the system.
The, ProCan framework allows both unsupervised and semi-supervised data analysis approaches, coupled with interactive and dynamic data visualization techniques, for dynamic biomarker discovery. Furthermore, when the ProCan system was tested against an established integromics framework (PICan), significant performance gains were observed in ProCan, even while preforming the same analytical format with a much larger data sample.
The ProCan system is a web based graphical user interface for managing and analysing integromics data. Its intuitive and dynamic nature renders it an effective, and efficient tool for conducting integromics research, allowing researchers to adopt efficient and reproducible analyses.