K-119. Absolute Comparison of Transcriptomic and Proteomic Data using Normalized Spectrum Counts to Estimate Protein Abundance

D. R. Johnson1, N. C. VerBerkmoes2, S. H. Zinder3, L. Alvarez-Cohen1,4;
1Univ. of California, Berkeley, CA, 2Oak Ridge Natl. Lab., Oak Ridge, TN, 3Cornell Univ., Ithaca, NY, 4Lawrence Berkeley Natl. Lab., Berkeley, CA.

Functional genomic techniques are typically used to measure the differential expression of individual genes. Although powerful, differential expression studies do not utilize all of the information available within functional genomic data sets. Most notable is that the absolute abundance levels of individual transcripts and/or proteins are not considered. Abundance information, however, can be used to compare the expression levels of different genes and provide a more complete picture of microbial physiology. We present a novel approach for comparing the abundance levels of different transcripts and proteins from functional genomic data sets for the bacterium Dehalococcoides ethenogenes. The transcriptome was measured by hybridizing total RNA and genomic DNA to replicate sets of Affymetrix microarrays. The RNA signal intensities were then normalized by the DNA signal intensities to control for differences in probe-set hybridization behavior. A comparison of the microarray data with RT-qPCR measurements confirmed that this approach could accurately estimate differences in transcript levels among different genes. The transcriptomic data was then directly compared with proteomic data generated by two-dimensional liquid chromatography coupled to tandem mass spectrometry (MS/MS). Protein abundance was estimated using two different quantification methods, spectral counts (SC) and normalized spectral counts (NSC). Spectral counts are a raw measure of the number of MS/MS spectra taken for each protein in a experiment, while normalized spectral counts take into account the length of each protein and the total number of spectra in the experiment We found that the NSC method generated protein abundances that had a higher correlation to measured transcript abundances than the SC method. Further analysis validated that the improved performance of the NSC method was attributed to measurements of exceptionally short or long peptides. Together, these results demonstrate that the comparison of absolute abundance levels of different transcripts and peptides is possible with only minor changes to existing functional genomic methodologies.