Dataset Information

Combination approaches improve predictive performance of diagnostic rules for mass-spectrometry proteomic data.

ABSTRACT: We consider a proteomic mass spectrometry case-control study for the construction of a diagnostic rule for patients' disease status allocation. We propose an approach for combining a collection of classifiers for the construction of a "combined" classification rule in order to enhance calibration and prediction ability. In a first stage this is achieved by building individual classifiers separately, each one using the entire proteomic data set. A double leave-one-out cross-validatory approach is used to estimate the class-predicted probabilities on which the combination method will be calibrated. The performance of the combination approach is examined both through a breast cancer proteomic data set and through simulation studies. Our experimental results indicate that in many circumstances gains in classification performance and predictive accuracy can be achieved.

SUBMITTER: Kakourou A

PROVIDER: S-EPMC4253302 | biostudies-literature | 2014 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Combination approaches improve predictive performance of diagnostic rules for mass-spectrometry proteomic data.

Kakourou Alexia A Vach Werner W Mertens Bart B

Journal of computational biology : a journal of computational molecular cell biology 20141201 12

We consider a proteomic mass spectrometry case-control study for the construction of a diagnostic rule for patients' disease status allocation. We propose an approach for combining a collection of classifiers for the construction of a "combined" classification rule in order to enhance calibration and prediction ability. In a first stage this is achieved by building individual classifiers separately, each one using the entire proteomic data set. A double leave-one-out cross-validatory approach is ...[more]

PMID: 25360568

Similar Datasets

Project description:Oral tumors, including highly invasive and metastatic oral melanoma (OM), non-tonsillar oral squamous cell carcinoma (OSCC) and benign tumors (BN), are common neoplasms in dogs. Although these tumors behave differently, limited data of their protein expression profiles have been exhibited, particularly at the proteome level. The present study aimed to i.) characterize peptide-mass fingerprints (PMFs) and identify potential protein candidates of OM, OSCC, BN and normal control subjects, using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) and liquid chromatography tandem mass spectrometry (LC-MS/MS), ii.) identify potential protein candidates associated with the diseases, using in-gel digestion coupled with mass spectrometric analysis (GeLC-MS/MS) and iii.) search for relationships between chemotherapy drugs and disease-perturbed proteins. A distinct cluster of each sample group and unique PMFs with identified protein candidates were revealed. The unique peptide fragment at 2,274 Da of sacsin molecular chaperone (SACS) was observed in early-stage OM whereas the fragment at 1,958 Da of sodium voltage-gated channel alpha subunit 10 (SCN10A) was presented in early- and late-stage OM. The peptide mass at 2,316 Da of Notch1 appeared in early-stage OM and benign oral tumors while the peptide mass at 2,505 Da of glutamate ionotropic receptor N-methyl-D-aspartate type subunit 3A (GRIN3A) was identified in all groups. Markedly expressed proteins from GeLC-MS/MS included Jumonji domain containing 1C (JMJD1C) in benign tumors, inversin (INVS) and rho guanine nucleotide exchange factor 28 (ARHGEF28) in OM, BTB domain-containing 16 (BTBD16) in OSCC, and protein tyrosine phosphatase non-receptor type 1 (PTPN1), BRCA2, DNA repair associated (BRCA2), WW domain binding protein 2 (WBP2), purinergic receptor P2Y1 and proteasome activator subunit 4 (PSME4) in all cancerous groups. The network connections between these proteins and chemotherapy drugs, cisplatin and doxorubicin, were also demonstrated. In conclusion, this study unveiled the unique PMFs and novel candidate protein markers of canine oral tumors.

Project description:IntroductionThe diagnosis of rheumatoid arthritis (RA) is based on a combined approach that includes serological markers such as rheumatoid factor (RF) and anti-citrullinated peptide/protein antibodies (ACPA). The goal of this study was to evaluate the clinical performance of several RF and ACPA immunoassays for the diagnosis of RA, as well as the diagnostic value of a combinatory approach with these markers.MethodsThe study cohort included 1,655 patients from the Swiss Clinical Quality Management registry with sera from 968 patients with RA and 687 disease controls, including patients with axial spondyloarthritis (n = 450) and psoriatic arthritis (n = 237). ACPA were determined by anti-CCP2 IgG enzyme-linked immunosorbent assay (ELISA), QUANTA Flash® CCP3 IgG [chemiluminescent immunoassay (CIA)], and QUANTA Lite® CCP3 IgG ELISA. RF was determined by ELISA (QUANTA Lite® RF IgM, RF IgA, and RF IgG) and with two research use only CIAs (QUANTA Flash® RF IgM and RF IgA).ResultsAll three ACPA assays showed good discrimination between RA patients and controls and good clinical performance. Overall, CCP3 performed better than CCP2. More pronounced differences were observed between the RF assays. We observed that CIA platforms for both RF IgM and RF IgA showed better performance than the ELISA platforms. Excellent and good total agreements were found between ELISA and CIA for CCP3 (total agreement 95.3%, kappa = 0.90), and between CCP2 and CCP3 ELISA (total agreement 86.6%, kappa = 0.73), respectively. RF IgM CIA and ELISA had a good qualitative agreement (86.5%, kappa = 0.73); RF IgA CIA and ELISA showed a moderate total agreement (78.5%, kappa = 0.53). When combinatory analyses were performed, the likelihood of RA increased with dual positivity and triple positivity and combining different markers resulted in higher odds ratio than the individual markers in all cases.ConclusionACPA and RF showed good clinical performance in this large Swiss cohort of RA patients and controls. Overall, the performance of CCP3 was superior to CCP2. The combination of these biomarkers in an interval model represents a potential tool for the diagnosis of RA patients.

Dataset Information

Combination approaches improve predictive performance of diagnostic rules for mass-spectrometry proteomic data.

Publications

Combination approaches improve predictive performance of diagnostic rules for mass-spectrometry proteomic data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets