Biomarkers, Outcomes and Stats Software (BOSS)
Although programmatic solutions are required to process and organize the huge data sets generated large studies and clinical trials, many research teams have smaller projects whose data they compile in conventional spreadsheet sofware (eg. Excel). To facilitate access to our analytical tools, we are building free, open-access, standalone modules for these tools. They are designed to allow investigators to input and analyze their data from a simple CSV (Excel) format. These tools are designed to run in the Google Chrome web browser. Our goal is build a platform that allows users to create interactive, living figures that can be shared and modified among collaborating research groups. The underlying code is open-source and our hope is that this will allow others to improve the existing modules and build new ones that enhance the functionality of the platform.
Interactive Visualization of Study Data
Using the BOSS analytical tools requires to files. First, a structured data set (CSV format) of the information to be visualized and analyzed. Second, a the visualization program (HTML file) that dynamically renders the structured data set into an interactive visualization. We currently have 2 tools available, both of which will read the same data set. The BOSS Relationships tool analyzes the correlations between biomarkers and outcomes, and analyzes their predictive reliability. The BOSS Relationships tool generates an interactive Scatter Chart. The BOSS Comparisons tool analyzes how parameters differ between cohorts of patients. The BOSS Comparisons tool generates an interactive Bar Chart. Use of these visualization does not require any special software, just a compatible web browser (Google Chrome). For more information on how to use these software tools refer to the Instructions, which are attached below.
BOSS Relationships Tool
BOSS Comparisons Tool
Software Environment and Open-Source Project
Notes on Statistical Analyses
For comparisons between groups, t-test (Student‰ŰŞs or Welch‰ŰŞs) or ANOVA are generally used to determine the significance of observed differences. For correlative analyses, simple linear regression (least squares) is used to generate the line of best fit (LOBF) and r2 value. To calculate significance, the probability that a value is predicted by the LOBF and the null hypothesis is estimated for each point in the data set. A t-test is used to determine the significance of the difference between the probabilities generated by the LOBF and those generated by the null hypothesis.
All visualizations are designed to run in Google Chrome. Other browsers are not supported and may not render properly.