Research & Teaching

banner image


The Fetrow research group has long been involved in understanding the details of molecular function of proteins. This problem is significant because of the ever-increasing numbers of protein sequences that are being determined through the genome sequencing projects. The vast majority of these proteins are of uncharacterized function; another large fraction of these proteins are mis-annotated at the molecular functional level. It is impossible, both methodologically and because of cost, to experimentally determine the function of every protein. Robust and automated computational approaches to this problem are essential.

We have developed a process, MISST (Harper, et al., 2017), to cluster proteins into functionally relevant or isofunctional groups.  The foundation of this work is a method called active site profiling (Cammer, et al. 2003).  As a proof-of-concept for functionally relevant clustering, active site profiling was applied manually to the identification of all members of each of the six subgroups of the peroxiredoxin superfamily, distinguishing all six functionally relevant groups (Prx1/AhpC, Prx6, PrxQ/BCP, Prx5, Tpx, and AhpE) that had previously been identified by experts (Nelson, et al. 2011), Even though mehods based on full sequence analysis are unable to identify each of these six groups. Active site profiling has also been applied to the cytochrome P450 family, resulting in comprehensive identification of functionally related proteins, subsequently validated experimentally (Gober, et al., 2016).

Iterative application of this profiling process has allowed us to demonstrate functionally relevant clustering on proteins of known structure (TuLIP, Knutson et al., 2017).  Additionally, MISST was developed to apply the iterative profiling concept to protein sequences for which structure is not known.  MISST, validated on the peroxiredoxins (Harper, et al., 2017), is the first process for functionally relevant clustering that is iterative and agglomerative, both adding sequences and identifying when a group should be subdivided into functionally relevant clusters.

MISST has been automated (Leutheauser and Fetrow, unpublished) and the automated process is being validated on a number of biologically important superfamilies, including the enolases, Radical SAMs, GSTs, beta lactamases and carbohydrate kinases.  In collaboration with Carol Parish’s laboratory and student Mikaela Rosen at the University of Richmond, we are creating a comprehensive atlas of arsenate reductases and functionally related phosphatases clustered into functionally relevant groups.

Our group also collaborate with Leslie Poole and Kim Nelson at Wake Forest University, experts in peroxiredoxin biology and biochemistry.  We have also collaborated with Eric Brustad at UNC Chapel Hill for the work on the cytochrome P450s.  Collaborators Patsy Babbitt, Tom Ferrin, and John “Scooter” Morris have been essential partners in the development of TuLIP and MISST.  The peroxiredoxin annotations identified in the 2011 Nelson paper are stored in the SFLD Database (see Akiva, et al. 2013RBVISFLD).

For more information on general research interests of our group or current projects we are involved in please visit the Research page.