Although genetic analysis can aid in disease management and improve patient care, the routine use of genetic information in a clinical environment is presently hindered by three main challenges:​ 1) the high cost of reagents, 2) the long experimental times, and 3) the demand for highly qualified labour. All three of these limitations and their underlying causes are studied and dealt with in this thesis by utilizing microfluidic devices.

This thesis describes the infrastructure developed for implementing and improving genetic analysis techniques on microfluidic devices. In particular, this infrastructure was applied to fluorescence in situ hybridization (FISH), a method based on the hybridization of fluorescently labelled probes to the chromosomal DNA in such a way that the resulting pattern of fluorescence indicates whether large-scale genetic rearrangements have taken place. Although useful for the detection, prognosis, and monitoring of many cancers, this important test is not routinely implemented due to its high cost.

The first issue with FISH that was addressed was reducing the amount of reagent consumed (probes are 3000-fold more expensive than gold). A microchip method was developed that efficiently localized expensive probe reagent over the cells and lowered the reagent cost-per-test by 20-fold.

The second issue with FISH was the long experimental time (days). To determine whether the underlying mechanism was diffusion or reaction limited, a mixing chip was designed that recirculated probe over the cells to overcome diffusion. It was determined that FISH was not diffusion limited (unlike DNA microarrays), and agitation approaches marginally improved the self-assembly process. The hybridization reaction is not easily improved with microchip methods;​ therefore to shorten experimental times, we used centromeric probes that hybridize in less than one hour.

The third issue with FISH was the demand for highly qualified labour (hours to days). The disjointed nature of the labour (minutes of activity separated by minutes of waiting) inefficiently utilizes skilled and expensive labour, and was addressed by integrating and automating the entire FISH protocol on a single microfluidic chip requiring only minutes of technician time.

Ultimately, this thesis work demonstrates the miniaturization infrastructure necessary to remove the above barriers for making FISH and many other self-assemblybased diagnostics more accessible, thereby enabling wide-spread implementation of important assays that can dramatically improve health-care delivery.