Experimental validation of methods for differential gene expression analysis and sample pooling in RNA-seq

BMC Genomics

Table 2 Validation of two pooling strategies for RNA-Seq

Parameters^a	Pooling 3 samples	Pooling 8 samples
Total number of identified DEGs^b	4175	2513
Sensitivity (True positivity rate) (%)	93.75	90.24
Specificity (True negativity rate) (%)	81.27	86.59
False positivity rate (%)	18.73	13.41
False negativity rate (%)	6.25	9.76
Positive predictive value (%)	0.36	2.94
Negative predictive value (%)	99.99	99.95
Agreement between identified DEGs^c	0.006	0.049
Correlation between reported LFC^d	0.380	0.517
Root-mean-square deviation of LFC^e	1.198	0.518

^aSequencing corresponding individual biological samples was the reference standard
^bDifferentially Expressed Genes (DEGs), after Benjamini-Hochberg false discovery correction at an expected rate of 5 %
^cInter-rater agreement Cohen’s kappa between sequencing individual samples (3 or 8/group) and sequencing pooled samples (3 or 8 biological replicates/pool; 2 pools/group) to identify DEGs
^dSpearman correlation coefficient between the logarithmic fold changes (LFC), which were estimated by sequencing individual samples and by sequencing pooled samples
^eStandard deviation of the differences between the LFC, estimated by sequencing individual samples and by sequencing pooled samples

ISSN: 1471-2164