This proposal is for CATTS, a feature learning technique optimized for use in multiplex mass spectrometry (MS) fingerprinting assays. MS fingerprints consist of a large number of chemical species, leading to very high dimensional feature spaces, and subsequent high false-discovery rates.
CATTS aims to reduce the size of this space, by using knowledge of the underlying biochemistry, as well as general-purpose clustering algorithms. Our preliminary results demonstrate that, when used as a feature-learning technique for a variety of classification methods, CATTS significantly improves assay sensitivity. This proposal takes our existing implementation of CATTS and extends it to support additional feature learning algorithms and classification methods. Additionally, its performance as a multiplex assay strategy will be tested on both protein and lipid MS fingerprint libraries, with an eye towards commercialization..
to public health: The detection of pathogens via mass spectroscopy fingerprinting is rapidly becoming a standard technique for clinical microbiology. However, high false detection rates and conflicting multiple identifications limit applicability, and make interpretation of results difficult. Our work on CATTS aims to improve the statistical performance of these assays. Preliminary results from studies on one dataset we intend to apply CATTS to suggest that UTIs and, in some cases antimicrobial resistance, can be detected, directly from patient samples. However, the statistical methods currently employed aren't reliable enough - the further development of CATTS will accelerate the development of this, and other mass-spectroscopy-based assays..
