Skip to main content

Table 3 Comparison results of various methods for SELC dataset on calculated log 2 fold change of gene expression

From: Improving RNA-Seq expression estimation by modeling isoform- and exon-specific read sequencing rate

Comparison Gene qRT-PCR Cufflinks RSEM MMSEQ NLDMseq
S vs. NS S100A8 +1.15 +1.13(0.02) +2.72(1.38) +3.20(1.78) +1.65(0.44)
  S100A9 +0.83 +1.69(1.03) +1.79(1.16) +1.86(1.24) +1.75(1.10)
  CYP4F2 +1.27 +4.12(2.24) +3.59(1.82) +9.51(6.49) +2.09(0.64)
C vs. NC CCL20 +4.01 +6.00(0.50) +4.66(0.16) +5.32(0.32) +7.08(0.76)
  IL8 +1.50 +1.51(0.01) +1.21(0.19) +1.67(0.11) +1.05(0.30)
  SCGB3A1 +0.51 +1.65(2.23) +1.76(2.45) +1.85(2.63) +1.82(2.57)
  SCGB1A1 −0.48 +1.37(3.89) +1.56(4.26) +1.02(3.13) +1.62(4.39)
Average AER NA 1.42 1.63 2.24 1.46
  1. The symbol “ +” stands for up-regulation and “ −” for down-regulation. Numbers in the brackets stand for absolute error rates (AER) compared with qRT-PCR results. AER is calculated by |(re)/r|, where |·| stands for absolute, and r and e represent qRT-PCR and RNA-seq measurements, respectively. The last line shows the average AER for each method