A Novel Group Testing Framework for Large-Scale Genotyping

The field of medical genetics has progressed to the point where it is possible to identify individuals who carry forms of genes known to cause debilitating genetic diseases, such as Huntington’s disease, cystic fibrosis, and Fragile X syndrome. A current research focus is determining people who are at risk for carrying these gene forms, which can be passed to their children and potentially expressed as a severe genetic disease.

Until recently, DNA sequencing systems used for this kind of diagnostics were limited to serial processing of a single specimen/region combination at a time. Now, massively parallel processing can read tens of millions of sequences in a single experimental run. Fully exploiting this technique would require “multiplexing” a large number of samples, which raises various technical and analytical difficulties, and it is this area that Professor Milenkovic will address during her Center appointment.

One promising multiplexing method developed at the Whitehead Institute uses compressive sensing methods to load pools of specimens in a “combinatorially controlled” manner that allows for subsequent determination of the set of mutation carriers. Professor Milenkovic plans to address shortcomings in current combinatorial schemes by developing an efficient semiquantitative group testing method that pools and tests groups of subjects simultaneously. The approach exploits the facts that individuals to be tested may be related, and consequently at correlated risk for being carriers, and that test subjects may have different gene copy number variations. If a test contains a number of affected individuals within a given precision range, the group is isolated and included in another round of testing. If the test result falls outside a precision range, the group is removed from future test panels. If the number of affected individuals is much smaller than the size of the population, one can expect to be able to reduce the number of required tests by orders of magnitude.

This project is the first known attempt to extend the theory of group testing to large-scale genotyping. Goals include determining the smallest number of tests required; new, constructive testing approaches; and the design of schemes that can be introduced and used in biological laboratories. Similar scenarios exist outside the field of genotyping, and the research results could have broad applications for theoretical computer science, mathematics, and bioengineering.