Folate system correlations in DNA microarray data

doi:10.1186/1471-2407-5-95

BMC Cancer
Volume 5

Viewing options:

Abstract
Full text
PDF (358KB)

Associated material:

Related literature:

Articles citing this article
on BioMed Central
on Google Scholar
on ISI Web of Science
on PubMed Central
Other articles by authors
on Google Scholar

Radivoyevitch T

on PubMed

Radivoyevitch T
Related articles/pages
on Google

on Google Scholar

on PubMed

Tools:

Post to:

Research article

Folate system correlations in DNA microarray data

Tomas Radivoyevitch

Department of Epidemiology and Biostatistics Case Western Reserve University, Cleveland, Ohio 44106 USA

author email corresponding author email

BMC Cancer 2005, 5:95doi:10.1186/1471-2407-5-95


Published:	4 August 2005

Abstract

Background

Gene expression data is abundantly available from the Gene Expression Omnibus (GEO) and various websites. Pathway specific analyses of gene-gene correlations across these datasets remain relatively unexplored, though they could be informative.

Methods

Folate gene expression data is explored here in two ways: (1) directly, using gene-gene scatter plots and gene expression time course plots; and (2) indirectly, using de novo purine synthesis (DNPS) and de novo thymidylate synthesis (DNTS) flux predictions of a folate model perturbed by relative gene expression modulations of its V_maxparameters.

Results

Positive correlations within and between the DNPS and DNTS folate cycles are observed in the folate gene expression data. For steady state measurements across childhood leukemia patients, positive correlations between DNPS and DNTS are consistent with higher proliferative fractions requiring higher levels of both fluxes. For cells exposed to ionizing radiation, transient increases in both pathways are consistent with DNA damage driven dNTP demand, and a steadily decreasing backdrop is consistent with radiation induced cell cycle arrest. By and large, folate model based flux predictions paralleled these findings, the main differences being a gain of correlation information for the TEL-AML1 leukemia data, and the loss of one interesting inference, namely, that RNA repair driven DNPS precedes DNA repair driven DNTS after a 10 gray dose of ionizing radiation.

Conclusion

Pathway focused correlation analyses of DNA microarray data can be informative, with or without a mathematical model. Conceptual models are essential. Mathematical model based analyses should supplement, but should not replace, direct data analyses.