Research article

Modelling and measuring single cell RNA expression levels find considerable transcriptional differences among phenotypically identical cells

Tatiana Subkhankulova , Michael J Gilchrist and Frederick J Livesey

Gurdon Institute and Department of Biochemistry, University of Cambridge, Tennis Court Road, Cambridge, CB2 1 QN, UK

author email corresponding author email

BMC Genomics 2008, 9:268doi:10.1186/1471-2164-9-268

Published:	3 June 2008

Abstract

Background

Phenotypically identical cells demonstrate predictable, robust behaviours. However, there is uncertainty as to whether phenotypically identical cells are equally similar at the underlying transcriptional level or if cellular systems are inherently noisy. To answer this question, it is essential to distinguish between technical noise and true variation in transcript levels. A critical issue is the contribution of sampling effects, introduced by the requirement to globally amplify the single cell mRNA population, to observed measurements of relative transcript abundance.

Results

We used single cell microarray data to develop simple mathematical models, ran Monte Carlo simulations of the impact of technical and sampling effects on single cell expression data, and compared these with experimental microarray data generated from single embryonic neural stem cells in vivo. We show that the actual distribution of measured gene expression ratios for pairs of neural stem cells is much broader than that predicted from our sampling effect model.

Conclusion

Our results confirm that significant differences in gene expression levels exist between phenotypically identical cells in vivo, and that these differences exceed any noise contribution from global mRNA amplification.