Dr. Mitch D Day

I was recently awarded an NSF post-doctoral fellowship in biological informatics. This is a three year award. Currently, I am spending the first year in Dr. Takashi Gojobori's lab group at the National Institute of Genetics in Mishima, Japan. After this experience, I will return to Idaho to work with Dr. James A Foster and the IBEST group at the University of Idaho, Moscow, ID.

During my dissertation work, I focused on probe design methods for microbial community analysis. For my in silico project, I tested for correlation between hybridization fingerprints generated from probes selected by certain algorithms and accepted microbial phylogenies. Here's the abstract from our planned publication of this work.

Abstract One of the major challenges facing the young field of metagenomics is the high level of diversity in the prokaryotic domains. Even though some individual bacterial or archaeal genomes may now be sequenced at relatively low cost, most uni-cellular organisms cannot be easily maintained in axenic culture and are only detectable by molecular signature, such as small sub-unit rRNA sequences. Probe-based detection methods, e.g. microarrays and bead-based techniques, often depend upon sequence data from closely- related organisms to select appropriate probes. This presents difficulties in the prokaryotic domains where new higher-level phyletic divisions are still routinely detected.

We hypothesized that a set of very short probes used to make a binary hybridization fingerprint can be used to identify previously unobserved sequences if the probe set returns a phylogenetic signal. We performed a series of simulation experiments on 177 bacterial genomes from the NCBI RefSeq database. We chose an information-theoretical measure for probe selection using the efficiency of partitioning of simulated shotgun libraries as the optimality criterion. Our studies show that the efficiency of partitioning of these probe sets declines consistently with increasing phylogenetic distance - as calculated using 16S rRNA marker sequences - from the training genome. We present these results and discuss potential applications of this finding to metagenomic profiling using bead-based techniques.