The Problem

For NIAID, RTI International conducted a study to help predict immune response to typhoid vaccine. The study, called the Population Genetics Project for Typhoid Vaccine (PopGen), recruited 4,000 participants at field sites in India. Blood and saliva samples were collected at baseline and over time after vaccination.  Samples were genotyped, used for proteomics, and measured for immune response factors. RTI subcontracted with DIFZ to bring insights in the data analysis phase of the project.

The Solution

RTI chose DIFZ’s GeneBrain® for the vaccine study data analysis because of the software’s ability to create advanced mathematical models of data from multiple diverse sources.

Our Computational Biology group applied our proprietary software, GeneBrain®, to speed analysis of genetics data for PopGen.  

GeneBrain® is the basis for a service that uses machine-learning algorithms on complex data. This software produces novel interactive 3D visualizations of data sets. It ranks and identifies optimum data feature sets that are predictive or characteristic of sample categories, such as “responsive” or “non-responsive” to vaccine.

The Benefits

GeneBrain® was able to:

  • Identify small groups of data features that are characteristic or predictive—even when single markers are not present. Traditional statistical analysis could not identify such markers.
  • Create a population genetics model that was highly successful in predicting high or low response based on a single nucleotide polymorphism (SNP) genotype alone.
  • Identify a small set of mixed type biomarkers that successfully characterized high or low immune response for the subject subset that had multiple data types, therefore suggesting the direction for further statistical analysis.
  • Explain other puzzling demographic findings.
  • Validate the information content of a new technique even at the lowest measurement levels.
  • Lead to the formulation and confirmation by testing of a new hypothesis. This demonstrated the usefulness of the intuitive visual models to scientists who conducted the laboratory research.