Nonlinear methods for discrimination and their application to classification of protein structures

MRS-based Metabolomics in Cancer Research

Metabolomics is a relatively new technique that is gaining importance very rapidly. MRS-based metabolomics, in particular, is becoming a useful tool in the study of body fluids, tissue biopsies and whole organisms. Advances in analytical techniques and data analysis methods have opened a new opportunity for such technology to contribute in the field of diagnostics. In the MRS approach to the diagnosis of disease, it is important that the analysis utilizes all the essential information in the spectra, is robust, and is non-subjective. Although some of the data analytic methods widely used in chemical and biological sciences are sketched, a more extensive discussion is given of a 5-stage Statistical Classification Strategy. This proposes powerful feature selection methods, based on, for example, genetic algorithms and novel projection techniques. The applications of MRS-based metabolomics in breast cancer, prostate cancer, colorectal cancer, pancreatic cancer, hepatobiliary cancers, gastric cancer, and brain cancer have been reviewed. While the majority of these applications relate to body fluids and tissue biopsies, some in vivo applications have also been included. It should be emphasized that the number of subjects studied must be sufficiently large to ensure a robust diagnostic classification. Before MRS-based metabolomics can become a widely used clinical tool, however, certain challenges need to be overcome. These include manufacturing user-friendly commercial instruments with all the essential features, and educating physicians and medical technologists in the acquisition, analysis, and interpretation of metabolomics data.

Creating robust, reliable, clinically relevant classifiers from spectroscopic data

I describe in detail the intimately connected feature extraction and classifier development stages of the data-driven Statistical Classification Strategy (SCS) and compare them with current practice used in MR spectroscopy. We initially created the SCS for the analysis of MR and IR spectra of biofluids and tissues, and subsequently extended it to analyze biomedical data in general. I focus on explaining how to extract discriminatory spectral features and create robust classifiers that can reliably discriminate diseases and disease states. I discuss our approach to identifying features that retain spectral identity and provisionally relate these features, averaged subregions of the spectra, to specific chemical entities ("metabolites"). Particular emphasis is placed on describing the steps required to help create classifiers whose accuracy doesn't deteriorate significantly when presented with new, unknown samples. A simple but powerful extension of the discovered features to detect metabolite-metabolite (feature-feature) interactions is also sketched. I contrast the advantages and disadvantages of using either spectral signatures or explicit metabolite concentrations derived from the spectra as sets of discriminatory features. At present, no clear-cut preference is obvious and more objective comparisons will be needed. Finally, I argue that clinical requirements and exigencies strongly suggest adopting a two-phase approach to diagnosis/prognosis. In the first phase the emphasis ought to be on providing as accurate a diagnosis as possible, without any attempt to identify "biomarkers." That should be the goal of the second, research phase, with a view of providing prognosis on disease progression.

Detecting colorectal cancer by 1H magnetic resonance spectroscopy of fecal extracts

Colorectal cancer is one of the most common cancers in the western world. Its early detection has been found to improve the prognosis of the patient, providing a wide window of opportunity for successful therapeutic interventions. However, current diagnostic techniques all have some limitations; there is a need to develop a better technique for routine screening purposes. We present a new methodology based on magnetic resonance spectroscopy of fecal extracts for the non-invasive detection of colorectal cancer. Five hundred twenty-three human subjects (412 with no colonic neoplasia and 111 with colorectal cancer, who were scheduled for colonoscopy or surgery) were recruited to donate a single sample of stool. One-dimensional (1)H magnetic resonance spectroscopy (MRS) experiments were performed on the supernatant of aqueous dispersions of the stool samples. Using a statistical classification strategy, several multivariate classifiers were developed. Applying the preprocessing, feature selection and classifier development stages of the Statistical Classification Strategy led to approximately 87% average balanced sensitivity and specificity for both training and monitoring sets, improving to approximately 92% when only crisp results, i.e. class assignment probabilities > or =75%, are considered. These results indicate that (1)H magnetic resonance spectroscopy of human fecal extracts, combined with appropriate data analysis methodology, has the potential to detect colorectal neoplasia accurately and reliably, and could be a useful addition to the current screening tools.

Protein sequence randomness and sequence/structure correlations

We investigated protein sequence/structure correlation by constructing a space of protein sequences, based on methods developed previously for constructing a space of protein structures. The space is constructed by using a representation of the amino acids as vectors of 10 property factors that encode almost all of their physical properties. Each sequence is represented by a distribution of overlapping sequence fragments. A distance between any two sequences can be calculated. By attaching a weight to each factor, intersequence distances can be varied. We optimize the correlation between corresponding distances in the sequence and structure spaces. The optimal correlation between the sequence and structure spaces is significantly better than that which results from correlating randomly generated sequences, having the overall composition of the data base, with the structure space. However, sets of randomly generated sequences, each of which approximates the composition of the real sequence it replaces, produce correlations with the structure space that are as good as that observed for the actual protein sequences. A connection is proposed with previous studies of the protein folding code. It is shown that the most important property factors for the correlation of the sequence and structure spaces are related to helix/bend preference, side chain bulk, and beta-structure preference.

Quantitative organization of the known protein x-ray structures. I. Methods and short-length-scale results

We address herein the problem of delineating the relationships between the known protein structures. In order to study this problem, methods have been developed to represent arbitrarily sized fragments of biopolymer backbone, and to compare distributions of such fragments. These methods are applied to a classification of 123 structures representing the entire set of known x-ray structures. The resulting data are analyzed (on the four-C alpha length scale) to determine both the large-scale organization of the set of known structures (i.e., the relationships between large groups of structures, each comprised of proteins that are structurally related) and its local structure (i.e., the quantitative degree of similarity between any two specific structures). It is shown that the set of structures forms a continuum of structural types, ranging from all-helical to all-sheet/barrel proteins. It is further demonstrated that the density of protein structures is not uniform across this continuum, but rather that structures cluster in certain regions, separated by regions of lower population. The properties of the various regions of the structural space are determined. The existence is demonstrated of strong quantitative correlations between the contents of different types of four-C alpha fragments within protein structures, which imply significant constraints on the types of architecture that can occur in proteins. Analysis of the distribution of structures demonstrates some hitherto unsuspected similarities and suggests that, in some circumstances, neither structural similarity nor sequence homology may be necessary conditions for evolutionary relationship between proteins. It is also suggested that these unsuspected similarities may imply similar folding mechanisms for structures of apparently different global architecture. Cases are also noted in which apparently similar structures may fold by different mechanisms. The connection between structure and dynamic properties is discussed, and a possible role of dynamics in the evolution of protein structures is suggested. The sensitivity of the methods presented herein to anomalies of structure refinement is demonstrated. It is suggested that the present results provide a framework for analyzing experimental results on structural similarity obtained using vibrational circular dichroism spectra, which are sensitive to local backbone structure.