3  Distributions

3.1 Histotype Distribution

Table 3.1: Histotype Distribution in Training Set by Processing Stage
Variable Levels CS1 CS2 CS3 Total
Selected Cohorts
Histotype HGSC 128 (44%) 655 (73%) 1808 (73%) 2591 (71%)
CCOC 48 (16%) 62 (7%) 164 (7%) 274 (7%)
ENOC 60 (20%) 49 (5%) 250 (10%) 359 (10%)
MUC 17 (6%) 58 (6%) 68 (3%) 143 (4%)
LGSC 19 (6%) 20 (2%) 36 (1%) 75 (2%)
Other 22 (7%) 59 (7%) 151 (6%) 232 (6%)
Total N (%) 294 (8%) 903 (25%) 2477 (67%) 3674 (100%)
QC
Histotype HGSC 122 (43%) 641 (73%) 1676 (74%) 2439 (71%)
CCOC 48 (17%) 62 (7%) 158 (7%) 268 (8%)
ENOC 60 (21%) 47 (5%) 213 (9%) 320 (9%)
MUC 16 (6%) 56 (6%) 65 (3%) 137 (4%)
LGSC 18 (6%) 20 (2%) 36 (2%) 74 (2%)
Other 22 (8%) 56 (6%) 125 (5%) 203 (6%)
Total N (%) 286 (8%) 882 (26%) 2273 (66%) 3441 (100%)
Main Histotypes
Histotype HGSC 122 (46%) 641 (78%) 1676 (78%) 2439 (75%)
CCOC 48 (18%) 62 (8%) 158 (7%) 268 (8%)
ENOC 60 (23%) 47 (6%) 213 (10%) 320 (10%)
MUC 16 (6%) 56 (7%) 65 (3%) 137 (4%)
LGSC 18 (7%) 20 (2%) 36 (2%) 74 (2%)
Total N (%) 264 (8%) 826 (26%) 2148 (66%) 3238 (100%)
Removed Duplicates
Histotype HGSC 118 (48%) 623 (78%) 1578 (78%) 2319 (76%)
CCOC 45 (18%) 56 (7%) 146 (7%) 247 (8%)
ENOC 56 (23%) 43 (5%) 200 (10%) 299 (10%)
MUC 13 (5%) 54 (7%) 55 (3%) 122 (4%)
LGSC 14 (6%) 19 (2%) 32 (2%) 65 (2%)
Total N (%) 246 (8%) 795 (26%) 2011 (66%) 3052 (100%)
Normalized and Recombined
Histotype HGSC 117 (49%) 622 (79%) 454 (97%) 1193 (79%)
CCOC 44 (18%) 55 (7%) 4 (1%) 103 (7%)
ENOC 55 (23%) 42 (5%) 4 (1%) 101 (7%)
MUC 12 (5%) 53 (7%) 4 (1%) 69 (5%)
LGSC 13 (5%) 18 (2%) 4 (1%) 35 (2%)
Total N (%) 241 (16%) 790 (53%) 470 (31%) 1501 (100%)
Removed Replicates
Histotype HGSC 9 (12%) 552 (78%) 454 (97%) 1015 (81%)
ENOC 38 (49%) 40 (6%) 4 (1%) 82 (7%)
CCOC 24 (31%) 53 (7%) 4 (1%) 81 (6%)
MUC 3 (4%) 50 (7%) 4 (1%) 57 (5%)
LGSC 3 (4%) 15 (2%) 4 (1%) 22 (2%)
Total N (%) 77 (6%) 710 (56%) 470 (37%) 1257 (100%)
Table 3.2: Histotype Distribution in Training, Confirmation, and Validation Sets
Variable Levels Training Confirmation Validation
Histotype HGSC 1015 (81%) 424 (66%) 699 (78%)
CCOC 81 (6%) 72 (11%) 69 (8%)
ENOC 82 (7%) 107 (17%) 88 (10%)
MUC 57 (5%) 27 (4%) 23 (3%)
LGSC 22 (2%) 12 (2%) 15 (2%)
Total N (%) 1257 (45%) 642 (23%) 894 (32%)

3.2 Cohort Distribution

Table 3.3: Pre-QC Cohort Distribution by CodeSet
CodeSet CS1
N = 294 		CS2
N = 903 		CS3
N = 2,477
Cohort
OOU 108 (37%) 43 (4.8%) 19 (0.8%)
OOUE 32 (11%) 30 (3.3%) 11 (0.4%)
VOA 145 (49%) 122 (14%) 538 (22%)
OVAR3 0 (0%) 150 (17%) 0 (0%)
OVAR11 0 (0%) 416 (46%) 0 (0%)
MAYO 6 (2.0%) 63 (7.0%) 0 (0%)
DOVE4 0 (0%) 0 (0%) 1,160 (47%)
TNCO 0 (0%) 0 (0%) 691 (28%)
MTL 3 (1.0%) 59 (6.5%) 0 (0%)
JAPAN 0 (0%) 8 (0.9%) 0 (0%)
POOL-CTRL 0 (0%) 12 (1.3%) 0 (0%)
POOL-1 0 (0%) 0 (0%) 31 (1.3%)
POOL-2 0 (0%) 0 (0%) 14 (0.6%)
POOL-3 0 (0%) 0 (0%) 13 (0.5%)
1 n (%)

3.3 Quality Control

3.3.1 Failed Samples

We use an aggregated QCFlag that considers a sample to have failed QC if any of the following QC conditions are flagged:

  • Linearity
  • Imaging
  • Smallest Positive Control
  • Normality
Table 3.4: Quality Control Summary
Quality Control Flag CS1
N = 294 		CS2
N = 903 		CS3
N = 2,477
Linearity
Failed 0 (0%) 4 (0.4%) 0 (0%)
Passed 294 (100%) 899 (100%) 2,477 (100%)
Imaging
Failed 3 (1.0%) 0 (0%) 4 (0.2%)
Passed 291 (99%) 903 (100%) 2,473 (100%)
Smallest Positive Control
Failed 0 (0%) 2 (0.2%) 0 (0%)
Passed 294 (100%) 901 (100%) 2,477 (100%)
Normality
Failed 5 (1.7%) 19 (2.1%) 200 (8.1%)
Passed 289 (98%) 884 (98%) 2,277 (92%)
Overall QC
Failed 8 (2.7%) 21 (2.3%) 204 (8.2%)
Passed 286 (97%) 882 (98%) 2,273 (92%)
1 n (%)

3.3.2 %GD vs. SNR

Figure 3.1: % Genes Detected vs. Signal to Noise Ratio
Figure 3.2: % Genes Detected vs. Signal to Noise Ratio (Zoomed)

3.4 Pairwise Gene Expression

Figure 3.3: Random1-Normalized CS1 vs. CS3 Gene Expression
Figure 3.4: Random1-Normalized CS2 vs. CS3 Gene Expression
Figure 3.5: HKgenes-Normalized CS1 vs. CS3 Gene Expression
Figure 3.6: HKgenes-Normalized CS2 vs. CS3 Gene Expression
Table 3.5: Wilcoxon signed rank test of gene correlations between normalization methods
Correlation Housekeeping Genes
N = 721		 Random1
N = 721		 p-value2
CS1 vs. CS3 0.84 (0.72, 0.90) 0.85 (0.64, 0.90) 0.160
CS2 vs. CS3 0.79 (0.69, 0.88) 0.84 (0.73, 0.90) <0.001
1 Median (Q1, Q3)
2 Wilcoxon signed rank test with continuity correction