Detection of cancer-specific markers amid massive mass spectral data

Abstract

We propose a comprehensive pattern recognition procedure that will achieve best discrimination between two or more sets of subjects with data in the same coordinate system. Applying the procedure to MS data of proteomic analysis of serum from ovarian cancer patients and serum from cancer-free individuals in the Food and Drug Administration͞National Cancer Institute Clinical Proteomics Database, we have achieved perfect discrimination (100% sensitivity, 100% specificity) of patients with ovarian cancer, including early-stage disease, from normal controls for two independent sets of data. Our procedure identifies the best subset of proteomic biomarkers for optimal discrimination between the groups and appears to have higher discriminatory power than other methods reported to date. For large-scale screening for diseases of relatively low prevalence such as ovarian cancer, almost perfect specificity and sensitivity of the detection system is critical to avoid unmanageably high numbers of false-positive cases. discriminant analysis ͉ random field ͉ resampling ͉ statgram

Classification of cancer types by measuring variants of host response proteins using SELDI serum assays

“…Use of protein markers and marker fragments in combination is likely to confer higher specificity than using any single one. [12][13][14][15] Singular value decomposition (SVD) was applied to the data of the 2 forms of ApoA1 and 4 forms of transthyretin to allow for visualization at reduced dimension. 16 The groups of ovarian cancer samples, controls and the breast, colon and prostate cancer samples each formed overlapping but distinguishable clusters with moderately good separation from one another in the 2-dimensional space spanned by the first and third components of the SVD analysis (Fig.…”

Serum proteome profiling detects myelodysplastic syndromes and identifies CXC chemokine ligands 4 and 7 as markers for advanced disease

“…Our profile distinguished MDS from non-MDS cytopenia patients independent of potential confounding by Hb, peripheral blood blast counts, age, and gender bias due to serum sampling by different hospitals or in vivo and in vitro hemolysis. Several reports have determined that class prediction results obtained from genomics or proteomics studies should be corroborated in independent sample sets (20)(21)(22)(23). Therefore, we tested our predictive profile in two separate independent validation sets run 5 months apart, rarely done before in proteomic or genomic studies (24)(25)(26).…”

Classification algorithms for phenotype prediction in genomics and proteomics

“…Zhu et al (58) used a statistical algorithm that can select a subset of k biomarkers from the marker list that could best discriminate between the groups in a training data set via the best k-subset discriminant method with high sensitivity and specificity.…”