Supplementary tables
Table S1. Five other singular enrichment tools analysed in this study
enrichment tool name / statistical method / annotation database§DAVID / Modified Fisher's Exact / GO; KEGG
GoMiner / Fisher's Exact / GO
Onto-express / Hypergeometric / GO
BINGO / Binomial / GO
GOseq / Wallenius / GO; KEGG
§For GO annotations, only the biological processes were analysed in this paper.
Table S2. Pe1 and Pe2 values for all datasets
dataset§ / background pairs# / positivea / negativeb / Pe1c / Pe2dBC / 8696535 / 2303612 / 1809833 / 0.26 / 0.21
CRA / 8696535 / 1624334 / 1435801 / 0.19 / 0.17
GC / 8696535 / 1383669 / 1131322 / 0.16 / 0.13
KIRC / 10339878 / 5766446 / 372039 / 0.56 / 0.04
LUAD / 10339878 / 5068853 / 73402 / 0.49 / 0.01
§BC, breast cancer; CRA, colorectal adenomas; GC, gastric cancer; KIRC, kidney renal clear cell carcinoma; LUAD, lung adenocarcinoma. #The number of all the pair-wise gene pairs of all the annotated gene. aThe number of positively associated gene pairs in the background gene pairs. bThe number of negatively associated gene pairs in the background gene pairs. cThe frequency of positively associated gene pairs in the background gene pairs. dThe frequency of negatively associated gene pairs in the background gene pairs.
Table S3. Correlated expression patterns of gene pairs with various types of functional links in the five datasets
dataset§ / link / positivea / negativeb / P-value1c / P-value2dBC / Inhibition (3012)# / 956 / 625 / 8.33×10-11 / 0.54
Phosphorylation (2523) / 831 / 466 / 4.23×10-13 / 0.99
Indirect effect (2053) / 787 / 308 / < 2.2×10-16 / 0.99
Ubiquitination (779) / 267 / 149 / 9.72×10-7 / 0.89
Dephosphorylation (646) / 202 / 124 / 3.80×10-3 / 0.86
missing interaction (134) / 42 / 43 / 0.12 / 1.50×10-3
Dissociation (127) / 55 / 23 / 3.13×10-5 / 0.80
state change (72) / 50 / 2 / 3.45×10-14 / 0.99
CRA / Inhibition (3012) / 699 / 441 / 3.17×10-10 / 0.99
Phosphorylation (2523) / 560 / 358 / 5.08×10-6 / 0.99
Indirect effect (2053) / 445 / 318 / 3.41×10-4 / 0.90
Ubiquitination (779) / 168 / 164 / 0.02 / 5.42×10-4
Dephosphorylation (646) / 138 / 119 / 0.05 / 0.11
missing interaction (134) / 33 / 18 / 0.05 / 0.86
Dissociation (127) / 40 / 20 / 3.66×10-4 / 0.63
state change (72) / 50 / 7 / < 2.2×10-16 / 0.96
GC / Inhibition (3012) / 669 / 350 / < 2.2×10-16 / 0.99
Phosphorylation (2523) / 575 / 288 / < 2.2×10-16 / 0.99
Indirect effect (2053) / 426 / 239 / 4.30×10-9 / 0.97
Ubiquitination (779) / 181 / 93 / 6.99×10-8 / 0.83
Dephosphorylation (646) / 106 / 96 / 0.38 / 0.09
missing interaction (134) / 31 / 30 / 0.02 / 1.90×10-3
Dissociation (127) / 39 / 9 / 2.35×10-5 / 0.99
state change (72) / 46 / 5 / < 2.2×10-16 / 0.97
KIRC / Inhibition (3593) / 2273 / 96 / < 2.2×10-16 / 0.99
Phosphorylation (2846) / 2015 / 80 / < 2.2×10-16 / 0.99
Indirect effect (2086) / 1592 / 29 / < 2.2×10-16 / 0.99
Ubiquitination (779) / 571 / 53 / < 2.2×10-16 / 1.12×10-5
Dephosphorylation (655) / 542 / 12 / < 2.2×10-16 / 0.99
missing interaction (134) / 116 / 1 / 1.82×10-14 / 0.99
Dissociation (129) / 107 / 1 / 5.67×10-11 / 0.99
state change (102) / 102 / 0 / < 2.2×10-16 / 1
LUAD / Inhibition (3593) / 1919 / 18 / 7.86×10-8 / 0.95
Phosphorylation (2846) / 1786 / 7 / < 2.2×10-16 / 0.99
Indirect effect (2086) / 1511 / 6 / < 2.2×10-16 / 0.99
Ubiquitination (779) / 536 / 5 / < 2.2×10-16 / 0.65
Dephosphorylation (655) / 469 / 4 / < 2.2×10-16 / 0.68
missing interaction (134) / 78 / 2 / 0.02 / 0.25
Dissociation (129) / 96 / 0 / 3.56×10-9 / 1
state change (102) / 99 / 0 / < 2.2×10-16 / 1
§BC, breast cancer; CRA, colorectal adenomas; GC, gastric cancer; KIRC, kidney renal clear cell carcinoma; LUAD, lung adenocarcinoma. #The number of the specific link. aThe number of positively associated gene pairs. bThe number of negatively associated gene pairs. cThe binomial P-value for positively associated gene pairs. dThe binomial P-value for negatively associated gene pairs.
Table S4. Correlation between the frequencies of the positively associated gene pairs and the degrees of imbalance between the up- and down-regulated genes in the pathways
BC / 0.39 / 4.32×10-9
CRA / 0.42 / 1.97×10-10
GC / 0.29 / 1.80×10-5
KIRC / 0.32 / 1.93×10-6
LUAD / 0.14 / 3.59×10-2
§BC, breast cancer; CRA, colorectal adenomas; GC, gastric cancer; KIRC, kidney renal clear cell carcinoma; LUAD, lung adenocarcinoma. aThe Spearman's correlation coefficient. bOne-sided P-value of Spearman's correlation.
Table S5. Results of random experiments
BC / 1 / 20 / 0.1152 / 0.0961 / 0.0000
CRA / 8 / 53 / 1.0608 / 0.0001 / 0.0000
GC / 5 / 41 / 0.6750 / 0.0009 / 0.0000
KIRC / 24 / 57 / 1.7291 / 0.0165 / 0.0000
LUAD / 4 / 44 / 0.6930 / 0.0000 / 0.0000
§BC, breast cancer; CRA, colorectal adenomas; GC, gastric cancer; KIRC, kidney renal clear cell carcinoma; LUAD, lung adenocarcinoma. aThe average number of pathways detected in 10,000 random experiments. bThe percentage of the random experiments in each of which the total number of detected pathways was not less than the observed number of pathways detected by the all-DE strategy in the 10,000 random experiments. cThe percentage of the random experiments in each of which the total number of detected pathways was not less than the observed number of pathways detected by the separate-DE strategy in 10,000 random experiments.
Table S6. - Multiple datasets used for evaluating the reproducibility of significant pathways for two tumour types
cancer§ / T:N# / data sourceBC / 176:10 / GSE22820
31:27 / GSE10810
183:13 / GSE5364
38:28 / GSE14548
54:12 / GSE29431
GC / 50:49 / GSE29998
134:134 / GSE29272
12:15 / GSE19826
65:19 / GSE13861
22:8 / GSE2685
§BC, breast cancer; GC, gastric cancer. #T denotes the number of tumour samples; N denotes the number of normal samples.
Table S7. - Number of significant gene sets from MSigDB detected using separate-DE and all-DE strategies
gene set / disease type§ / up DEa / down DEb / all DEc / number1d / number2eC1 / BC / 10 / 2 / 0 / 12 / 12
CRA / 0 / 0 / 0 / - / -
GC / 2 / 0 / 0 / 2 / 2
KIRC / 8 / 4 / 0 / 12 / 12
LUAD / 9 / 3 / 4 / 9 / 8
C3 / BC / 20 / 214 / 13 / 223 / 223
CRA / 46 / 151 / 7 / 191 / 191
GC / 66 / 25 / 19 / 72 / 72
KIRC / 12 / 12 / 5 / 21 / 21
LUAD / 11 / 107 / 19 / 104 / 104
C4 / BC / 156 / 63 / 59 / 163 / 159
CRA / 238 / 111 / 178 / 175 / 175
GC / 246 / 105 / 134 / 217 / 216
KIRC / 174 / 32 / 96 / 123 / 115
LUAD / 185 / 87 / 140 / 138 / 134
C6 / BC / 12 / 37 / 21 / 35 / 35
CRA / 24 / 26 / 30 / 25 / 25
GC / 39 / 32 / 6 / 65 / 65
KIRC / 19 / 8 / 35 / 12 / 12
LUAD / 14 / 23 / 44 / 17 / 17
C7 / BC / 519 / 238 / 863 / 198 / 183
CRA / 617 / 542 / 1215 / 162 / 159
GC / 1054 / 14 / 226 / 842 / 833
KIRC / 140 / 1 / 24 / 126 / 125
LUAD / 128 / 170 / 71 / 241 / 241
§BC, breast cancer; CRA, colorectal adenomas; GC, gastric cancer; KIRC, kidney renal clear cell carcinoma; LUAD, lung adenocarcinoma. aThe number of significant gene sets detected by analysing the up-regulated genes. bThe number of significant gene sets detected by analysing the down-regulated genes. cThe number of significant gene sets detected by analysing all DE genes. dThe number of significant gene sets detected by applying the separate-DE strategy but missed by applying the all-DE strategy. eThe number of pathways detected by applying the separate-DE but missed by applying the all-DE strategy that have significantly imbalanced numbers of up- and down-regulated genes (P-value < 0.05, Fisher’s exact test).