Skip to main content

Table 4 Performance metrics on differentially expressed (DE) plasmodes

From: Inferential considerations for low-count RNA-seq transcripts: a case study on the dominant prairie grass Andropogon gerardii

edgeR robust DF = 50

edgeR robust DF = 10

edgeR robust DF = 4

edgeR robust \( \mathrm{D}\mathrm{F} = \widehat{DF} \)

edgeR classic \( \mathrm{D}\mathrm{F} = \widehat{DF} \)

DESEQ2

Power - All transcripts

0.6495 a

0.6275 b

0.6215 b

0.6229 b

0.5704 c

0.5418 d

(0.00610)

(0.00610)

(0.00463)

(0.00463)

(0.00435)

(0.00435)

Power - Low-count transcripts

0.3922 a

0.3692 b

0.3657 b

0.3677 b

0.2647 c

0.1785 d

(0.00120)

(0.00120)

(0.00010)

(0.00010)

(0.00008)

(0.00008)

Precision - All transcripts

0.3607 a

0.4393 b

0.5218 c

0.5323 c

0.6321 d

0.6586 e

(0.00792)

(0.01153)

(0.01192)

(0.01114)

(0.00589)

(0.00371)

Precision - Low-count transcripts

0.1800 a

0.2289 b

0.2904 c

0.2963 c

0.3605 d

0.3915 e

(0.00811)

(0.01305)

(0.01432)

(0.01378)

(0.01363)

(0.01727)

NPV - All transcripts

0.9917 a

0.9915 a

0.9915 a

0.9915 a

0.9902 b

0.9891 c

(0.00016)

(0.00016)

(0.00016)

(0.00016)

(0.00016)

(0.00016)

NPV - Low-count transcripts

0.9917 a

0.9915 a

0.9914 a

0.9915 a

0.9902 b

0.9891 c

(0.00016)

(0.00016)

(0.00016)

(0.00016)

(0.00016)

(0.00016)

Accuracy - All transcripts

0.9718 a

0.9778 b

0.9821 b

0.9826 c

0.9858 d

0.9862 e

(0.00080)

(0.00080)

(0.00024)

(0.00024)

(0.00012)

(0.00014)

Accuracy - Low-count transcripts

0.9679 a

0.9744 b

0.9794 c

0.9798 c

0.9841 d

0.9855 e

(0.00122)

(0.00122)

(0.00079)

(0.00079)

(0.00028)

(0.00028)

FPR - All transcripts

0.0221 a

0.0155 b

0.011 b

0.0106 c

0.0063 d

0.0053 e

(0.00086)

(0.00086)

(0.00024)

(0.00024)

(0.00001)

(0.00010)

FPR - Low-count transcripts

0.0244 a

0.0175 b

0.0124 c

0.0121 c

0.0063 d

0.0037 e

(0.00128)

(0.00128)

(0.00082)

(0.00082)

(0.00019)

(0.00019)

  1. Least square mean estimates (and corresponding SEM, shown in parentheses) for true positive rate (TPR; i.e. power), positive predictive value (PPV; i.e. precision), negative predictive value (NPV), accuracy and false positive rate (FPR) for differential expression at FDR = 0.05 on all transcripts and on low-count transcripts yielded by DESeq2, EdgeR classic or EdgeR robust, implemented on DE plasmodes of RNA-seq data. a,b,c,d,e, indicate significant differences (Tukey-Kramer adjusted P < 0.05) within a row