Open Access Highly Accessed Open Badges Research article

A comparison of RNA amplification techniques at sub-nanogram input concentration

Julie E Lang15*, Mark Jesus M Magbanua2, Janet H Scott2, G Mike Makrigiorgos4, Gang Wang4, Scot Federman2, Laura J Esserman1, John W Park2 and Christopher M Haqq23

Author Affiliations

1 Department of Surgery, UCSF Comprehensive Cancer Center, 1500 Divisadero Street, San Francisco, CA 94143, USA

2 Department of Medical Oncology, UCSF Comprehensive Cancer Center, San Francisco, CA 94143, USA

3 Department of Urology, UCSF Comprehensive Cancer Center, San Francisco, CA 94143, USA

4 Dana Farber Cancer Institute, Harvard Medical School, 44 Binney Street, Boston, MA 02115, USA

5 Department of Surgery, Arizona Cancer Center, University of Arizona, 1515 N, Campbell Ave #1968, PO Box 245024, Tucson, AZ 85724, USA

For all author emails, please log on.

BMC Genomics 2009, 10:326  doi:10.1186/1471-2164-10-326

Published: 20 July 2009



Gene expression profiling of small numbers of cells requires high-fidelity amplification of sub-nanogram amounts of RNA. Several methods for RNA amplification are available; however, there has been little consideration of the accuracy of these methods when working with very low-input quantities of RNA as is often required with rare clinical samples. Starting with 250 picograms-3.3 nanograms of total RNA, we compared two linear amplification methods 1) modified T7 and 2) Arcturus RiboAmp HS and a logarithmic amplification, 3) Balanced PCR. Microarray data from each amplification method were validated against quantitative real-time PCR (QPCR) for 37 genes.


For high intensity spots, mean Pearson correlations were quite acceptable for both total RNA and low-input quantities amplified with each of the 3 methods. Microarray filtering and data processing has an important effect on the correlation coefficient results generated by each method. Arrays derived from total RNA had higher Pearson's correlations than did arrays derived from amplified RNA when considering the entire unprocessed dataset, however, when considering a gene set of high signal intensity, the amplified arrays had superior correlation coefficients than did the total RNA arrays.


Gene expression arrays can be obtained with sub-nanogram input of total RNA. High intensity spots showed better correlation on array-array analysis than did unfiltered data, however, QPCR validated the accuracy of gene expression array profiling from low-input quantities of RNA with all 3 amplification techniques. RNA amplification and expression analysis at the sub-nanogram input level is both feasible and accurate if data processing is used to focus attention to high intensity genes for microarrays or if QPCR is used as a gold standard for validation.