Skip to main content
. 2018 May 16;9:176. doi: 10.3389/fgene.2018.00176

Table 5.

Partitioning contribution statistics by likelihood ratio tests of eight selected biomarkers via bootstrap forest modeling of sample classification under seven inferred MOA clusters with reference gene lists at different tiers of statistical stringency (mode-of-action training RNAseq dataset, TGxSEQC; GEO accession number: GSE55347).

Gene Symbol 1,510 DEGREEs 65 Profiler DEGREEs 8 Biomarkers Statistic
Ucp3 8.4178 14.2830 14.5505 G2
(9.89%) (17.12%) (22.92%) (G2gene/G2total, %)
1 1 1 Rank by G2 [highest: 1]
0.2091 0.0266 0.0241 p-value [df: MOA Clusters-1]
Tmem86b 0.0000 2.4276 8.5938
(0.00%) (2.91%) (13.54%)
1363 6 2
1.0000 0.8765 0.1977
Hadhb 0.3465 1.4197 8.3181
(0.41%) (1.70%) (13.11%)
83 20 3
0.9992 0.9647 0.2157
Gsdmd 0.0000 0.1950 7.5273
(0.00%) (0.23%) (11.86%)
603 50 4
1.0000 0.9999 0.2748
Acaa1b 0.0000 2.3052 7.3439
(0.00%) (2.76%) (11.57%)
142 9 5
1.0000 0.8896 0.2902
Acaa1a 0.4146 2.1847 6.5803
(0.49%) (2.62%) (10.37%)
61 12 6
0.9987 0.9020 0.3614
Tfam 0.0000 2.2218 6.3974
(0.00%) (2.66%) (10.08%)
1331 10 7
1.0000 0.8982 0.3802
Sugct 0.0000 0.0000 4.1599
(0.00%) (0.00%) (6.55%)
1308 56 8
1.0000 1.0000 0.6551