A unique tool offering single and multi-omics biomarker discovery with a combination of statistical, machine learning and biological network methods.
InSyBio Biomarkers tool offers a patent pending pipeline for the discovery of predictive biomarkers, biosignatures using single, multi-omics data and/or clinical information.
The computational biomarker discovery solution of InSyBio biomarkers, is using advanced multi- objective ensemble dimensionality reduction techniques combined with information about the interactions of molecules (genes, RNA, proteins) in order to be able to identify a minimum number of biomarkers that should be selected to predict the phenotype under study with the highest accuracy that can be achieved. A minimal number of samples (10 per phenotype under study) are required to allow for the application of this pipeline. The benefits of the overall approach are maximized when combined with InSyBio BioNets network comparison module.
InSyBio biomarkers pipeline can achieve an average increase of 10% in predictive accuracy of the retrieved biomarkers and models. Moreover, it achieves a decrease of at least 80% of the overall time and cost of the process by decreasing the needs to acquire additional licenses for other products and outsource parts of the analysis and by reducing the requirements for validation experiments since the number of retrieved potential biomarkers is reduced significantly.
InSyBio Biomarkers key properties include:
- Data preprocessing using a combination of normalization, filtering, scaling and missing values imputation techniques
- Semi-automatic statistical analysis: automatically select the best statistical test for your dataset
- Bioinformatics visualization of your omics data with heatmaps, volcano plots, boxplots and excel-like tabular files
- Identification of multiple biomarker solutions and training machine learning predictors with them
- Using trained machine learning models and identified multiple biomarkers to predict the phenotype on new data
