#################################################
### Elastic net + GLM ###
#################################################
# save p-values from each test
save(stage2.glmPois.pval, file=paste(npop,"n_stage2_pvalue_glmPois.RData",sep=""))
save(stage2.glmNB.pval, file=paste(npop,"n_stage2_pvalue_glmNB.RData",sep=""))
save(stage2.glmquasiPois.pval, file=paste(npop,"n_stage2_pvalue_glmquasiPois.RData",sep=""))
# save detected features from each test
save(stage1.sig, file=paste(npop,"n_stage1_sig.RData",sep=""))
save(stage2.glmPois.sig, file=paste(npop,"n_stage2_sig_glmPois.RData",sep=""))
save(stage2.glmNB.sig, file=paste(npop,"n_stage2_sig_glmNB.RData",sep=""))
save(stage2.glmquasiPois.sig, file=paste(npop,"n_stage2_sig_glmquasiPois.RData",sep=""))
# save results from variable selection
save(stage1.perf, file="varsel_perf.RData")
# save results from GLM-Poisson (after running elastic net)
save(stage2.glmPois.perf, file="varsel_glmPois_perf.RData")
save(stage2.glmquasiPois.perf, file="varsel_glmquasiPois_perf.RData")
save(stage2.glmNB.perf, file="varsel_glmNB_perf.RData")
### The performance of Stage1 (Elastic net) ###
> stage1.perf [[1]] [[1]]$npop [1] 10 [[1]]$perf sens spec acc err simid1 0.19 0.8644 0.797 0.203 simid2 0.23 0.8667 0.803 0.197 simid3 0.03 0.9844 0.889 0.111 [[1]]$perf.avg sens spec acc err 0.1500000 0.9051667 0.8296667 0.1703333 [[1]]$power 1.5 2 5 7 10 25 50 75 90 100 simid1 10 0 30 30 10 10 10 40 30 20 simid2 30 20 10 20 10 40 20 40 20 20 simid3 20 0 0 0 0 0 0 0 10 0 [[1]]$power.avg 1.5 2 5 7 10 20.000000 6.666667 13.333333 16.666667 6.666667 25 50 75 90 100 16.666667 10.000000 26.666667 20.000000 13.333333 [[2]] [[2]]$npop [1] 25 [[2]]$perf sens spec acc err simid1 0.30 0.8711 0.814 0.186 simid2 0.24 0.9156 0.848 0.152 simid3 0.67 0.5667 0.577 0.423 [[2]]$perf.avg sens spec acc err 0.4033333 0.7844667 0.7463333 0.2536667 [[2]]$power 1.5 2 5 7 10 25 50 75 90 100 simid1 10 10 40 50 30 30 60 20 30 20 simid2 20 20 30 30 30 10 20 30 40 10 simid3 30 40 40 60 50 90 90 90 100 80 [[2]]$power.avg 1.5 2 5 7 10 25 20.00000 23.33333 36.66667 46.66667 36.66667 43.33333 50 75 90 100 56.66667 46.66667 56.66667 36.66667 [[3]] [[3]]$npop [1] 50 [[3]]$perf sens spec acc err simid1 0.72 0.6989 0.701 0.299 simid2 0.85 0.4811 0.518 0.482 simid3 0.80 0.6144 0.633 0.367 [[3]]$perf.avg sens spec acc err 0.7900000 0.5981333 0.6173333 0.3826667 [[3]]$power 1.5 2 5 7 10 25 50 75 90 100 simid1 20 50 70 60 80 80 90 90 100 80 simid2 40 40 90 90 90 100 100 100 100 100 simid3 30 70 70 60 90 90 100 100 100 90 [[3]]$power.avg 1.5 2 5 7 10 30.00000 53.33333 76.66667 70.00000 86.66667 25 50 75 90 100 90.00000 96.66667 96.66667 100.00000 90.00000
### The performance of Stage2 (GLM - Neg. Bin.) ###
> stage2.glmNB.perf
[[1]]
[[1]]$npop
[1] 10
[[1]]$perf
sens spec acc err
simid1 0.13 0.9822 0.897 0.103
simid2 0.16 0.9756 0.894 0.106
simid3 0.01 0.9956 0.897 0.103
[[1]]$perf.avg
sens spec acc err
0.1000000 0.9844667 0.8960000 0.1040000
[[1]]$power
1.5 2 5 7 10 25 50 75 90 100
simid1 0 0 20 20 0 0 10 40 30 10
simid2 10 10 0 20 10 30 10 30 20 20
simid3 0 0 0 0 0 0 0 0 10 0
[[1]]$power.avg
1.5 2 5 7 10
3.333333 3.333333 6.666667 13.333333 3.333333
25 50 75 90 100
10.000000 6.666667 23.333333 20.000000 10.000000
[[2]]
[[2]]$npop
[1] 25
[[2]]$perf
sens spec acc err
simid1 0.27 0.9367 0.870 0.130
simid2 0.21 0.9622 0.887 0.113
simid3 0.49 0.9089 0.867 0.133
[[2]]$perf.avg
sens spec acc err
0.3233333 0.9359333 0.8746667 0.1253333
[[2]]$power
1.5 2 5 7 10 25 50 75 90 100
simid1 10 10 40 30 30 30 50 20 30 20
simid2 20 10 30 30 10 10 20 30 40 10
simid3 10 20 20 30 30 50 80 80 100 70
[[2]]$power.avg
1.5 2 5 7 10 25
13.33333 13.33333 30.00000 30.00000 23.33333 30.00000
50 75 90 100
50.00000 43.33333 56.66667 33.33333
[[3]]
[[3]]$npop
[1] 50
[[3]]$perf
sens spec acc err
simid1 0.53 0.9467 0.905 0.095
simid2 0.60 0.9556 0.920 0.080
simid3 0.54 0.9667 0.924 0.076
[[3]]$perf.avg
sens spec acc err
0.55666667 0.95633333 0.91633333 0.08366667
[[3]]$power
1.5 2 5 7 10 25 50 75 90 100
simid1 0 0 20 30 50 70 90 90 100 80
simid2 0 0 20 50 30 100 100 100 100 100
simid3 0 0 30 20 30 80 100 100 100 80
[[3]]$power.avg
1.5 2 5 7 10
0.00000 0.00000 23.33333 33.33333 36.66667
25 50 75 90 100
83.33333 96.66667 96.66667 100.00000 86.66667
###########################################################
### 2. GLM-Poisson Model ###
###########################################################
# save p-values from GLM-Pois test
save(glmPois.pval, file=paste(npop,"n_pvalue_glmPois.RData",sep=""))
# save detected features from GLM-Pois test
save(glmPois.sig, file=paste(npop,"n_sig_glmPois.RData",sep=""))
> save(glmPois.perf, file="glmPois_perf.RData")
> glmPois.perf
### 3. GLM-NB Model ###
###########################################################
# save p-values from GLM-NB test
save(glmNB.pval, file=paste(npop,"n_pvalue_glmNB.RData",sep=""))
# save detected features from GLM-NB test
save(glmNB.sig, file=paste(npop,"n_stage2_sig_glmNB.RData",sep=""))
> save(glmNB.perf, file="glmNB_perf.RData") > glmNB.perf
############################################################## 4. GLM-Quasi Poisson Model ###
###########################################################
# save p-values from GLM-Quasi Poisson test
save(glmquasiPois.pval, file=paste(npop,"n_pvalue_glmquasiPois.RData",sep=""))
# save detected features from GLM-Quasi Poisson test
save(glmquasiPois.sig, file=paste(npop,"n_sig_glmquasiPois.RData",sep=""))
> save(glmquasiPois.perf, file="glmPois_perf.RData") > glmquasiPois.perf
##############################################################
### 5. metagenomeSeq (using CSS Normalization) ###
##############################################################
# save p-values
save(metagenomeSeqCSS.pval, file=paste(npop,"n_pvalue_metagenomeSeqCSS.RData",sep=""))
# save detected features
save(metagenomeSeqCSS.sig, file=paste(npop,"n_sig_metagenomeSeqCSS.RData",sep=""))
> save(metagenomeSeqCSS.perf, file="metagenomeSeqCSS_perf.RData")> metagenomeSeqCSS.perf
No comments:
Post a Comment