Histogram of PPDE frequencies for all genes, i.e. no FDR cutoff.
Histogram of PPDE frequencies for all genes with FDR=0.01 cutoff.
Heatmap of gene counts for genes with the top 50 posterior FC values. Relapse samples: 5080, 6655, 6709, 8825, 8830, 8831, 9121, 1108724. Disease-free samples: 6491, 6503, 6504, 6529, 6588, 6705, 6706, 6787, 7332.
How cool is this?! There's definitely segregation between the relapse and disease-free patients! The next three heatmaps show similar segregation.
Heatmap of gene counts for genes with the top 50 real FC values. Relapse samples: 5080, 6655, 6709, 8825, 8830, 8831, 9121, 1108724. Disease-free samples: 6491, 6503, 6504, 6529, 6588, 6705, 6706, 6787, 7332.
Heatmap of gene counts for genes with the bottom 50 posterior FC values. Relapse samples: 5080, 6655, 6709, 8825, 8830, 8831, 9121, 1108724. Disease-free samples: 6491, 6503, 6504, 6529, 6588, 6705, 6706, 6787, 7332.
Heatmap of gene counts for genes with the bottom 50 real FC values. Relapse samples: 5080, 6655, 6709, 8825, 8830, 8831, 9121, 1108724. Disease-free samples: 6491, 6503, 6504, 6529, 6588, 6705, 6706, 6787, 7332.
Heatmap of gene counts for top 50 most variables, organized by cluster instead of individual sample. DF = disease-free. R = relapse.
Heatmap of gene counts for top 500 most variables, organized by cluster instead of individual sample. DF = disease-free. R = relapse.
Heatmap of gene counts for the top 50 most variable genes with FDR=0.01 cutoff. The results are really all over the place, i.e. there's not a clear segregation between the relapse and disease-free samples. Relapse samples: 5080, 6655, 6709, 8825, 8830, 8831, 9121, 1108724. Disease-free samples: 6491, 6503, 6504, 6529, 6588, 6705, 6706, 6787, 7332.
Same thing going on here, only for the top 500 variable genes with FDR=0.01 cutoff. Relapse samples: 5080, 6655, 6709, 8825, 8830, 8831, 9121, 1108724. Disease-free samples: 6491, 6503, 6504, 6529, 6588, 6705, 6706, 6787, 7332.
Heatmap of gene counts for the top 50 most variable genes WITHOUT a FDR cutoff. DF = disease-free, R = relapse.
Heatmap of gene counts for the top 500 most variable genes WITHOUT a FDR cutoff. DF = disease-free, R = relapse
Volcano plot of PPDE vs. real FC values for genes with FDR=0.01 cutoff. Left plot takes -log10(PPDE), while the right plot is just PPDE. Blue = PPDE > 0.95 & abs(log2(RealFC)) < 15; Red = PPDE > 0.95 & abs(log2(RealFC)) < 5.
Same as previous volcano plots, but with the posterior FC values instead of the real FC values. Blue = PPDE > 0.95 & abs(log2(PostFC)) < 15; Red = PPDE > 0.95 & abs(log2(PostFC)) < 5.
These four volcano plots are the same as the volcano plots above them. I just used ggplot here instead of the default plot function to better show the overlap of the plotted points. Blue = abs(log2(RealFC or PostFC)) > 5 & PPDE > 0.95); Red = every other data point.
These eight plots are similar to the ones before them, except I transformed the PPDE value to a FDR value. This posed some problems while using ggplot and the only workaround that I found seems to slightly modify with distribution. For the first four plots, Blue = PPDE > 0.95 & abs(log2(PostFC or RealFC)) < 15; Red = PPDE > 0.95 & abs(log2(PostFC or realFC)) < 5. For the last four plots, Blue = abs(log2(RealFC or PostFC)) > 5 & PPDE > 0.95); Red = every other data point.
## PPEE PPDE PostFC RealFC C1Mean
## ENSG00000079689 0.000000e+00 1 0.0001148164 9.637593e-05 0.5779245
## ENSG00000125820 0.000000e+00 1 0.0024883120 1.708478e-04 0.0000000
## ENSG00000242389 0.000000e+00 1 0.0031383683 2.148726e-04 0.0000000
## ENSG00000237957 0.000000e+00 1 0.0031805948 2.173413e-04 0.0000000
## ENSG00000276690 0.000000e+00 1 0.0032904136 2.242667e-04 0.0000000
## ENSG00000226023 0.000000e+00 1 0.0037703730 2.592758e-04 0.0000000
## ENSG00000261771 0.000000e+00 1 0.0045800636 3.017141e-04 0.0000000
## ENSG00000250657 0.000000e+00 1 0.0050727581 3.433257e-04 0.0000000
## ENSG00000251357 0.000000e+00 1 0.0057402447 3.831778e-04 0.0000000
## ENSG00000172238 0.000000e+00 1 0.0063910917 4.299200e-04 0.0000000
## ENSG00000205916 0.000000e+00 1 0.0068595916 4.679862e-04 0.0000000
## ENSG00000204279 0.000000e+00 1 0.0075736572 5.199130e-04 0.0000000
## ENSG00000168267 0.000000e+00 1 0.0077556924 5.231635e-04 0.0000000
## ENSG00000281955 0.000000e+00 1 0.0083311386 5.610780e-04 0.0000000
## ENSG00000264404 0.000000e+00 1 0.0082532316 5.675037e-04 0.0000000
## ENSG00000249196 0.000000e+00 1 0.0084951974 5.787549e-04 0.0000000
## ENSG00000275113 0.000000e+00 1 0.0009754241 6.507598e-04 0.2624113
## ENSG00000231824 0.000000e+00 1 0.0105119591 7.248111e-04 0.0000000
## ENSG00000255551 2.553513e-15 1 0.0108116117 7.415305e-04 0.0000000
## ENSG00000172023 0.000000e+00 1 0.0118631725 8.214034e-04 0.0000000
## C2Mean
## ENSG00000079689 6100.31508
## ENSG00000125820 58.52161
## ENSG00000242389 46.52921
## ENSG00000237957 46.00059
## ENSG00000276690 44.57977
## ENSG00000226023 38.55897
## ENSG00000261771 33.13396
## ENSG00000250657 29.11686
## ENSG00000251357 26.08755
## ENSG00000172238 23.25014
## ENSG00000205916 21.35815
## ENSG00000204279 19.22399
## ENSG00000168267 19.10449
## ENSG00000281955 17.81283
## ENSG00000264404 17.61103
## ENSG00000249196 17.26847
## ENSG00000275113 418.59507
## ENSG00000231824 13.78670
## ENSG00000255551 13.47562
## ENSG00000172023 12.16429
Top 20 differentially expressed genes with highest real fold change. Note, I still need to convert the ensembl IDs to hgnc symbols.