Blind spots of quantitative RNA-seq: the limits for assessing abundance, differential expression, and isoform switching

doi:10.1186/1471-2105-14-370

PMC full text:

BMC Bioinformatics. 2013; 14: 370.

Published online 2013 Dec 24. doi: 10.1186/1471-2105-14-370

Copyright/LicenseRequest permission to reuse: Copyright © 2013 Rehrauer et al.; licensee BioMed Central Ltd.
This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Table 4

Area under the curve for the differential expression

Alignment and count method	Exact test	Log ratio	T-test-log
STAR-countOverlaps	0.917	0.9191	0.8845
tophat-countOverlaps	0.9162	0.9183	0.8837
STAR-htseq	0.9119	0.916	0.8797
tophat-htseq	0.9109	0.9151	0.8784
RSEM-TPM	0.9264	0.9342	0.8964
RSEM-posteriorTPM	0.9273	0.9355	0.897
RSEM-count	0.9295	0.9328	0.8959
RSEM-posteriorCount	0.9303	0.9338	0.8965

The table shows the overall AUC value for the detection of differential expression. The samples were designed such that each gene was differentially expressed in multiple comparisons. Using different types of input we apply 3 testing scores: exactTest (edgeR package), log-ratio (as an example for effect-size estimation), t-test on the log-expression scores (as a baseline in order to contrast the benefit of using a count-model).