Related Experiment Videos
A study of inter-lab and inter-platform agreement of DNA microarray data
Huixia Wang1, Xuming He, Mark Band
1Department of Statistics, University of Illinois at Urbana-Champaign, 101 Illini Hall, 725 South Wright Street, Champaign, Illinois 61820, USA. hwang22@uiuc.edu
BMC Genomics
|May 13, 2005
Summary
Comparing gene expression data across labs and platforms shows acceptable agreement. However, lab effects significantly impact data consistency more than platform differences.
Area of Science:
- Genomics
- Bioinformatics
- Biostatistics
Background:
- Gene expression profile data from DNA microarrays are rapidly accumulating.
- Comparing data across different laboratories and technology platforms presents significant challenges due to complexity and variability.
Purpose of the Study:
- To assess the inter-laboratory and inter-platform agreement of microarray data.
- To quantify the degree of consistency achievable across different labs and technologies.
Main Methods:
- Utilized statistical measures including Pearson correlation, intraclass correlation, and kappa coefficients.
- Compared data from three distinct platforms: Affymetrix GeneChip, custom cDNA arrays, and custom oligo arrays.
- Evaluated agreement within and across three different laboratories.
Main Results:
- Technology platforms demonstrated acceptable agreement when compared to within-platform variability.
- Agreement between different technologies within the same lab exceeded agreement between the same technologies in different labs.
- Lab effects, particularly when confounded with RNA sample effects, had a greater impact on data agreement than platform effects.
Conclusions:
- While microarray platforms show reasonable agreement, laboratory-specific factors significantly influence data consistency.
- High consistency among replicates enhances agreement across technologies and labs.
- Minimizing lab effects is crucial for improving the reliability of cross-platform and cross-laboratory microarray data comparisons.