Skip to main content
. 2009 May 4;25(13):1647–1654. doi: 10.1093/bioinformatics/btp288

Table 1.

Comparison of ALLENMINER results to published sets of regionally enriched genes

Benchmark set Num ROC Optimal
genes AUC Score TPR FPR
Manually curated ABA listsa
Cerebellum 99 0.977 3.553 0.970 0.050
Cortex 98 0.979 1.715 0.969 0.063
Hippocampus 98 0.961 3.515 0.908 0.072
HY 95 0.977 1.530 0.968 0.070
Midbrain 95 0.951 1.352 0.926 0.081
Medulla 95 0.953 2.460 0.958 0.112
OLF 99 0.953 2.489 0.949 0.111
Pallidum 98 0.965 1.299 0.949 0.088
Pons 97 0.951 1.580 0.969 0.103
Retrohippocampal region 99 0.958 1.819 0.929 0.089
Striatum 97 0.991 2.514 0.959 0.019
Thalamus 99 0.982 2.553 0.960 0.045

Published microarray results
Amygdalab 20 0.606 1.478 0.600 0.348
Cerebellum 112 0.679 1.037 0.652 0.369
Hippocampus 53 0.699 1.441 0.660 0.368
Olfactory bulb 68 0.735 1.371 0.647 0.275
Periaqueductal gray 47 0.679 0.609 0.617 0.344

VMHc 37 0.792 1.412 0.703 0.185

ROC analysis quantified the accuracy of the ALLENMINER enrichment score in discriminating regionally enriched genes, as defined by manual curation or microarray analysis, from the remaining ABA genes. The ‘optimal’ true positive (TPR) and false positive rates (FPR) for the score threshold closest to the perfect classifier (TPR = 1, FPR = 0), along with the AUC, are described for each benchmark set.

aLein et al. (2007).

bZirlinger et al. (2001).

cKurrasch et al. (2007).