Correlation between molecular cancer subtypes and selected clinical features

Overview

Introduction

This pipeline computes the correlation between cancer subtypes identified by different molecular patterns and selected clinical features.

Summary

Testing the association between subtypes identified by 10 different clustering approaches and 8 clinical features across 502 patients, 35 significant findings detected with P value < 0.05 and Q value < 0.25.

CNMF clustering analysis on array-based mRNA expression data identified 3 subtypes that do not correlate to any clinical features.
Consensus hierarchical clustering analysis on array-based mRNA expression data identified 3 subtypes that do not correlate to any clinical features.
3 subtypes identified in current cancer cohort by 'CN CNMF'. These subtypes correlate to 'Time to Death', 'PATHOLOGY.T', 'PATHOLOGICSPREAD(M)', and 'TUMOR.STAGE'.
3 subtypes identified in current cancer cohort by 'METHLYATION CNMF'. These subtypes correlate to 'Time to Death', 'GENDER', 'PATHOLOGY.T', 'PATHOLOGICSPREAD(M)', and 'TUMOR.STAGE'.
CNMF clustering analysis on RPPA data identified 3 subtypes that correlate to 'Time to Death', 'PATHOLOGY.T', 'PATHOLOGICSPREAD(M)', and 'TUMOR.STAGE'.
Consensus hierarchical clustering analysis on RPPA data identified 3 subtypes that correlate to 'Time to Death', 'PATHOLOGY.T', 'PATHOLOGICSPREAD(M)', and 'TUMOR.STAGE'.
CNMF clustering analysis on sequencing-based mRNA expression data identified 3 subtypes that correlate to 'Time to Death', 'GENDER', 'PATHOLOGY.T', 'PATHOLOGICSPREAD(M)', and 'TUMOR.STAGE'.
Consensus hierarchical clustering analysis on sequencing-based mRNA expression data identified 3 subtypes that correlate to 'Time to Death', 'GENDER', 'PATHOLOGY.T', 'PATHOLOGY.N', 'PATHOLOGICSPREAD(M)', and 'TUMOR.STAGE'.
CNMF clustering analysis on sequencing-based miR expression data identified 3 subtypes that correlate to 'Time to Death', 'GENDER', 'PATHOLOGY.T', 'PATHOLOGICSPREAD(M)', and 'TUMOR.STAGE'.
Consensus hierarchical clustering analysis on sequencing-based miR expression data identified 3 subtypes that correlate to 'Time to Death' and 'PATHOLOGY.T'.

Results

Overview of the results

Table 1. Get Full Table Overview of the association between subtypes identified by 10 different clustering approaches and 8 clinical features. Shown in the table are P values (Q values). Thresholded by P value < 0.05 and Q value < 0.25, 35 significant findings detected.


Clinical Features	Time to Death	AGE	GENDER	KARNOFSKY PERFORMANCE SCORE	PATHOLOGY T	PATHOLOGY N	PATHOLOGICSPREAD(M)	TUMOR STAGE
Statistical Tests	logrank test	ANOVA	Fisher's exact test	ANOVA	Chi-square test	Fisher's exact test	Fisher's exact test	Chi-square test
mRNA CNMF subtypes	0.231 (1.00)	0.795 (1.00)	0.634 (1.00)		0.00911 (0.357)	0.0704 (1.00)	0.12 (1.00)	0.0134 (0.482)
mRNA cHierClus subtypes	0.422 (1.00)	0.607 (1.00)	0.82 (1.00)		0.00779 (0.327)	0.124 (1.00)	0.0988 (1.00)	0.0102 (0.388)
CN CNMF	0.000165 (0.00922)	0.0782 (1.00)	0.177 (1.00)	0.889 (1.00)	0.00196 (0.0923)	0.0272 (0.924)	0.000486 (0.0253)	0.00154 (0.074)
METHLYATION CNMF	6.32e-06 (0.000392)	0.00858 (0.352)	0.00326 (0.147)	0.134 (1.00)	2.64e-10 (2.01e-08)	0.147 (1.00)	6.41e-05 (0.00378)	3.89e-09 (2.92e-07)
RPPA CNMF subtypes	4.15e-07 (2.9e-05)	0.127 (1.00)	0.22 (1.00)	0.552 (1.00)	4.3e-06 (0.000279)	0.149 (1.00)	5.29e-06 (0.000333)	9.71e-08 (7.09e-06)
RPPA cHierClus subtypes	7.29e-07 (5.03e-05)	0.0772 (1.00)	0.336 (1.00)	0.0291 (0.961)	1.02e-07 (7.34e-06)	0.252 (1.00)	0.000361 (0.0191)	1.59e-07 (1.13e-05)
RNAseq CNMF subtypes	1.63e-06 (0.000109)	0.09 (1.00)	0.000154 (0.00878)	0.66 (1.00)	4.85e-06 (0.00031)	0.0117 (0.433)	0.000645 (0.0326)	1.27e-06 (8.64e-05)
RNAseq cHierClus subtypes	3.34e-08 (2.47e-06)	0.358 (1.00)	0.00326 (0.147)	0.357 (1.00)	7.56e-12 (5.82e-10)	0.000966 (0.0474)	2e-06 (0.000132)	2.63e-12 (2.05e-10)
MIRseq CNMF subtypes	8.12e-06 (0.000495)	0.0537 (1.00)	0.00265 (0.122)	0.149 (1.00)	0.00028 (0.0154)	0.00693 (0.298)	0.000305 (0.0164)	8.11e-05 (0.0047)
MIRseq cHierClus subtypes	2.74e-05 (0.00165)	0.0521 (1.00)	0.182 (1.00)	0.849 (1.00)	0.00064 (0.0326)	0.00892 (0.357)	0.282 (1.00)	0.0143 (0.501)

Clustering Approach #1: 'mRNA CNMF subtypes'

Table S1. Get Full Table Description of clustering approach #1: 'mRNA CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	34	24	14

'mRNA CNMF subtypes' versus 'Time to Death'

P value = 0.231 (logrank test), Q value = 1

Table S2. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	71	13	0.5 - 101.1 (32.6)
subtype1	33	4	0.5 - 101.1 (31.0)
subtype2	24	8	0.5 - 93.3 (36.7)
subtype3	14	1	1.3 - 84.4 (25.0)

Figure S1. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'mRNA CNMF subtypes' versus 'AGE'

P value = 0.795 (ANOVA), Q value = 1

Table S3. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	71	60.5 (12.4)
subtype1	33	60.2 (13.8)
subtype2	24	59.9 (11.1)
subtype3	14	62.6 (11.3)

Figure S2. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #2: 'AGE'

'mRNA CNMF subtypes' versus 'GENDER'

P value = 0.634 (Fisher's exact test), Q value = 1

Table S4. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	29	43
subtype1	15	19
subtype2	10	14
subtype3	4	10

Figure S3. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #3: 'GENDER'

'mRNA CNMF subtypes' versus 'PATHOLOGY.T'

P value = 0.00911 (Chi-square test), Q value = 0.36

Table S5. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3
ALL	41	14	17
subtype1	23	4	7
subtype2	10	4	10
subtype3	8	6	0

Figure S4. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

'mRNA CNMF subtypes' versus 'PATHOLOGY.N'

P value = 0.0704 (Fisher's exact test), Q value = 1

Table S6. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	35	3
subtype1	18	0
subtype2	10	3
subtype3	7	0

Figure S5. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

'mRNA CNMF subtypes' versus 'PATHOLOGICSPREAD(M)'

P value = 0.12 (Fisher's exact test), Q value = 1

Table S7. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	67	5
subtype1	33	1
subtype2	20	4
subtype3	14	0

Figure S6. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'mRNA CNMF subtypes' versus 'TUMOR.STAGE'

P value = 0.0134 (Chi-square test), Q value = 0.48

Table S8. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	40	15	12	5
subtype1	23	5	5	1
subtype2	9	4	7	4
subtype3	8	6	0	0

Figure S7. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

Clustering Approach #2: 'mRNA cHierClus subtypes'

Table S9. Get Full Table Description of clustering approach #2: 'mRNA cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	15	23	34

'mRNA cHierClus subtypes' versus 'Time to Death'

P value = 0.422 (logrank test), Q value = 1

Table S10. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	71	13	0.5 - 101.1 (32.6)
subtype1	15	2	1.3 - 84.4 (24.2)
subtype2	23	7	0.5 - 93.3 (36.8)
subtype3	33	4	0.5 - 101.1 (31.0)

Figure S8. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'mRNA cHierClus subtypes' versus 'AGE'

P value = 0.607 (ANOVA), Q value = 1

Table S11. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	71	60.5 (12.4)
subtype1	15	63.2 (11.2)
subtype2	23	59.1 (10.7)
subtype3	33	60.4 (14.0)

Figure S9. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'mRNA cHierClus subtypes' versus 'GENDER'

P value = 0.82 (Fisher's exact test), Q value = 1

Table S12. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	29	43
subtype1	5	10
subtype2	9	14
subtype3	15	19

Figure S10. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

'mRNA cHierClus subtypes' versus 'PATHOLOGY.T'

P value = 0.00779 (Chi-square test), Q value = 0.33

Table S13. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3
ALL	41	14	17
subtype1	9	6	0
subtype2	9	4	10
subtype3	23	4	7

Figure S11. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

'mRNA cHierClus subtypes' versus 'PATHOLOGY.N'

P value = 0.124 (Fisher's exact test), Q value = 1

Table S14. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	35	3
subtype1	7	0
subtype2	11	3
subtype3	17	0

Figure S12. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

'mRNA cHierClus subtypes' versus 'PATHOLOGICSPREAD(M)'

P value = 0.0988 (Fisher's exact test), Q value = 1

Table S15. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	67	5
subtype1	15	0
subtype2	19	4
subtype3	33	1

Figure S13. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'mRNA cHierClus subtypes' versus 'TUMOR.STAGE'

P value = 0.0102 (Chi-square test), Q value = 0.39

Table S16. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	40	15	12	5
subtype1	9	6	0	0
subtype2	8	4	7	4
subtype3	23	5	5	1

Figure S14. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

Clustering Approach #3: 'CN CNMF'

Table S17. Get Full Table Description of clustering approach #3: 'CN CNMF'

Cluster Labels	1	2	3
Number of samples	132	197	164

'CN CNMF' versus 'Time to Death'

P value = 0.000165 (logrank test), Q value = 0.0092

Table S18. Clustering Approach #3: 'CN CNMF' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	490	158	0.1 - 111.0 (35.2)
subtype1	132	50	0.2 - 97.5 (36.6)
subtype2	195	43	0.1 - 111.0 (37.5)
subtype3	163	65	0.1 - 109.9 (28.5)

Figure S15. Get High-res Image Clustering Approach #3: 'CN CNMF' versus Clinical Feature #1: 'Time to Death'

'CN CNMF' versus 'AGE'

P value = 0.0782 (ANOVA), Q value = 1

Table S19. Clustering Approach #3: 'CN CNMF' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	492	60.6 (12.2)
subtype1	131	62.5 (11.8)
subtype2	197	59.5 (12.7)
subtype3	164	60.3 (11.7)

Figure S16. Get High-res Image Clustering Approach #3: 'CN CNMF' versus Clinical Feature #2: 'AGE'

'CN CNMF' versus 'GENDER'

P value = 0.177 (Fisher's exact test), Q value = 1

Table S20. Clustering Approach #3: 'CN CNMF' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	171	322
subtype1	42	90
subtype2	78	119
subtype3	51	113

Figure S17. Get High-res Image Clustering Approach #3: 'CN CNMF' versus Clinical Feature #3: 'GENDER'

'CN CNMF' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.889 (ANOVA), Q value = 1

Table S21. Clustering Approach #3: 'CN CNMF' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	36	88.3 (23.0)
subtype1	14	88.6 (26.0)
subtype2	9	91.1 (10.5)
subtype3	13	86.2 (26.9)

Figure S18. Get High-res Image Clustering Approach #3: 'CN CNMF' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'CN CNMF' versus 'PATHOLOGY.T'

P value = 0.00196 (Chi-square test), Q value = 0.092

Table S22. Clustering Approach #3: 'CN CNMF' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3	T4
ALL	242	64	176	11
subtype1	54	19	57	2
subtype2	116	25	55	1
subtype3	72	20	64	8

Figure S19. Get High-res Image Clustering Approach #3: 'CN CNMF' versus Clinical Feature #5: 'PATHOLOGY.T'

'CN CNMF' versus 'PATHOLOGY.N'

P value = 0.0272 (Fisher's exact test), Q value = 0.92

Table S23. Clustering Approach #3: 'CN CNMF' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	228	18
subtype1	55	5
subtype2	91	2
subtype3	82	11

Figure S20. Get High-res Image Clustering Approach #3: 'CN CNMF' versus Clinical Feature #6: 'PATHOLOGY.N'

'CN CNMF' versus 'PATHOLOGICSPREAD(M)'

P value = 0.000486 (Fisher's exact test), Q value = 0.025

Table S24. Clustering Approach #3: 'CN CNMF' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	417	76
subtype1	102	30
subtype2	181	16
subtype3	134	30

Figure S21. Get High-res Image Clustering Approach #3: 'CN CNMF' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'CN CNMF' versus 'TUMOR.STAGE'

P value = 0.00154 (Chi-square test), Q value = 0.074

Table S25. Clustering Approach #3: 'CN CNMF' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	238	52	123	80
subtype1	52	16	33	31
subtype2	116	19	45	17
subtype3	70	17	45	32

Figure S22. Get High-res Image Clustering Approach #3: 'CN CNMF' versus Clinical Feature #8: 'TUMOR.STAGE'

Clustering Approach #4: 'METHLYATION CNMF'

Table S26. Get Full Table Description of clustering approach #4: 'METHLYATION CNMF'

Cluster Labels	1	2	3
Number of samples	97	120	66

'METHLYATION CNMF' versus 'Time to Death'

P value = 6.32e-06 (logrank test), Q value = 0.00039

Table S27. Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	281	95	0.1 - 109.9 (28.5)
subtype1	96	49	0.2 - 84.7 (28.6)
subtype2	119	21	0.1 - 109.6 (31.5)
subtype3	66	25	0.1 - 109.9 (21.3)

Figure S23. Get High-res Image Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #1: 'Time to Death'

'METHLYATION CNMF' versus 'AGE'

P value = 0.00858 (ANOVA), Q value = 0.35

Table S28. Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	283	61.5 (12.0)
subtype1	97	63.6 (10.2)
subtype2	120	59.0 (12.9)
subtype3	66	63.0 (11.9)

Figure S24. Get High-res Image Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #2: 'AGE'

'METHLYATION CNMF' versus 'GENDER'

P value = 0.00326 (Fisher's exact test), Q value = 0.15

Table S29. Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	96	187
subtype1	21	76
subtype2	52	68
subtype3	23	43

Figure S25. Get High-res Image Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #3: 'GENDER'

'METHLYATION CNMF' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.134 (ANOVA), Q value = 1

Table S30. Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	28	92.5 (8.0)
subtype1	6	88.3 (7.5)
subtype2	17	92.4 (8.3)
subtype3	5	98.0 (4.5)

Figure S26. Get High-res Image Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'METHLYATION CNMF' versus 'PATHOLOGY.T'

P value = 2.64e-10 (Chi-square test), Q value = 2e-08

Table S31. Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3	T4
ALL	132	36	107	8
subtype1	20	14	59	4
subtype2	78	18	24	0
subtype3	34	4	24	4

Figure S27. Get High-res Image Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #5: 'PATHOLOGY.T'

'METHLYATION CNMF' versus 'PATHOLOGY.N'

P value = 0.147 (Fisher's exact test), Q value = 1

Table S32. Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	127	9
subtype1	43	5
subtype2	54	1
subtype3	30	3

Figure S28. Get High-res Image Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #6: 'PATHOLOGY.N'

'METHLYATION CNMF' versus 'PATHOLOGICSPREAD(M)'

P value = 6.41e-05 (Fisher's exact test), Q value = 0.0038

Table S33. Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	232	51
subtype1	66	31
subtype2	109	11
subtype3	57	9

Figure S29. Get High-res Image Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'METHLYATION CNMF' versus 'TUMOR.STAGE'

P value = 3.89e-09 (Chi-square test), Q value = 2.9e-07

Table S34. Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	130	24	73	56
subtype1	20	9	35	33
subtype2	78	12	19	11
subtype3	32	3	19	12

Figure S30. Get High-res Image Clustering Approach #4: 'METHLYATION CNMF' versus Clinical Feature #8: 'TUMOR.STAGE'

Clustering Approach #5: 'RPPA CNMF subtypes'

Table S35. Get Full Table Description of clustering approach #5: 'RPPA CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	160	150	144

'RPPA CNMF subtypes' versus 'Time to Death'

P value = 4.15e-07 (logrank test), Q value = 2.9e-05

Table S36. Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	452	151	0.1 - 111.0 (34.3)
subtype1	158	75	0.2 - 97.5 (30.6)
subtype2	150	29	0.3 - 111.0 (38.3)
subtype3	144	47	0.1 - 96.8 (30.2)

Figure S31. Get High-res Image Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'RPPA CNMF subtypes' versus 'AGE'

P value = 0.127 (ANOVA), Q value = 1

Table S37. Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	453	60.4 (12.3)
subtype1	160	61.9 (12.1)
subtype2	149	60.2 (12.7)
subtype3	144	59.0 (12.0)

Figure S32. Get High-res Image Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #2: 'AGE'

'RPPA CNMF subtypes' versus 'GENDER'

P value = 0.22 (Fisher's exact test), Q value = 1

Table S38. Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	151	303
subtype1	45	115
subtype2	55	95
subtype3	51	93

Figure S33. Get High-res Image Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #3: 'GENDER'

'RPPA CNMF subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.552 (ANOVA), Q value = 1

Table S39. Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	34	93.5 (7.7)
subtype1	9	91.1 (7.8)
subtype2	15	94.7 (8.3)
subtype3	10	94.0 (7.0)

Figure S34. Get High-res Image Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'RPPA CNMF subtypes' versus 'PATHOLOGY.T'

P value = 4.3e-06 (Chi-square test), Q value = 0.00028

Table S40. Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3	T4
ALL	224	54	165	11
subtype1	51	24	77	8
subtype2	93	13	43	1
subtype3	80	17	45	2

Figure S35. Get High-res Image Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

'RPPA CNMF subtypes' versus 'PATHOLOGY.N'

P value = 0.149 (Fisher's exact test), Q value = 1

Table S41. Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	208	16
subtype1	77	10
subtype2	60	2
subtype3	71	4

Figure S36. Get High-res Image Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

'RPPA CNMF subtypes' versus 'PATHOLOGICSPREAD(M)'

P value = 5.29e-06 (Fisher's exact test), Q value = 0.00033

Table S42. Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	380	74
subtype1	116	44
subtype2	139	11
subtype3	125	19

Figure S37. Get High-res Image Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'RPPA CNMF subtypes' versus 'TUMOR.STAGE'

P value = 9.71e-08 (Chi-square test), Q value = 7.1e-06

Table S43. Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	219	43	114	78
subtype1	49	18	46	47
subtype2	92	10	37	11
subtype3	78	15	31	20

Figure S38. Get High-res Image Clustering Approach #5: 'RPPA CNMF subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

Clustering Approach #6: 'RPPA cHierClus subtypes'

Table S44. Get Full Table Description of clustering approach #6: 'RPPA cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	132	215	107

'RPPA cHierClus subtypes' versus 'Time to Death'

P value = 7.29e-07 (logrank test), Q value = 5e-05

Table S45. Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	452	151	0.1 - 111.0 (34.3)
subtype1	132	43	0.2 - 111.0 (36.6)
subtype2	215	53	0.1 - 96.8 (37.0)
subtype3	105	55	0.1 - 91.4 (21.2)

Figure S39. Get High-res Image Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'RPPA cHierClus subtypes' versus 'AGE'

P value = 0.0772 (ANOVA), Q value = 1

Table S46. Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	453	60.4 (12.3)
subtype1	131	61.9 (12.0)
subtype2	215	59.1 (12.9)
subtype3	107	61.4 (11.3)

Figure S40. Get High-res Image Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'RPPA cHierClus subtypes' versus 'GENDER'

P value = 0.336 (Fisher's exact test), Q value = 1

Table S47. Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	151	303
subtype1	39	93
subtype2	79	136
subtype3	33	74

Figure S41. Get High-res Image Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

'RPPA cHierClus subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.0291 (ANOVA), Q value = 0.96

Table S48. Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	34	93.5 (7.7)
subtype1	10	89.0 (8.8)
subtype2	18	96.7 (4.9)
subtype3	6	91.7 (9.8)

Figure S42. Get High-res Image Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'RPPA cHierClus subtypes' versus 'PATHOLOGY.T'

P value = 1.02e-07 (Chi-square test), Q value = 7.3e-06

Table S49. Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3	T4
ALL	224	54	165	11
subtype1	55	19	57	1
subtype2	135	20	58	2
subtype3	34	15	50	8

Figure S43. Get High-res Image Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

'RPPA cHierClus subtypes' versus 'PATHOLOGY.N'

P value = 0.252 (Fisher's exact test), Q value = 1

Table S50. Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	208	16
subtype1	66	4
subtype2	82	4
subtype3	60	8

Figure S44. Get High-res Image Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

'RPPA cHierClus subtypes' versus 'PATHOLOGICSPREAD(M)'

P value = 0.000361 (Fisher's exact test), Q value = 0.019

Table S51. Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	380	74
subtype1	113	19
subtype2	191	24
subtype3	76	31

Figure S45. Get High-res Image Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'RPPA cHierClus subtypes' versus 'TUMOR.STAGE'

P value = 1.59e-07 (Chi-square test), Q value = 1.1e-05

Table S52. Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	219	43	114	78
subtype1	53	15	43	21
subtype2	133	17	42	23
subtype3	33	11	29	34

Figure S46. Get High-res Image Clustering Approach #6: 'RPPA cHierClus subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

Clustering Approach #7: 'RNAseq CNMF subtypes'

Table S53. Get Full Table Description of clustering approach #7: 'RNAseq CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	199	180	101

'RNAseq CNMF subtypes' versus 'Time to Death'

P value = 1.63e-06 (logrank test), Q value = 0.00011

Table S54. Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	478	155	0.1 - 111.0 (34.3)
subtype1	199	44	0.1 - 111.0 (37.0)
subtype2	179	83	0.1 - 90.3 (30.6)
subtype3	100	28	0.1 - 93.3 (35.2)

Figure S47. Get High-res Image Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq CNMF subtypes' versus 'AGE'

P value = 0.09 (ANOVA), Q value = 1

Table S55. Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	479	60.6 (12.2)
subtype1	198	61.7 (12.2)
subtype2	180	60.6 (11.8)
subtype3	101	58.4 (12.7)

Figure S48. Get High-res Image Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #2: 'AGE'

'RNAseq CNMF subtypes' versus 'GENDER'

P value = 0.000154 (Fisher's exact test), Q value = 0.0088

Table S56. Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	167	313
subtype1	90	109
subtype2	45	135
subtype3	32	69

Figure S49. Get High-res Image Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'GENDER'

'RNAseq CNMF subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.66 (ANOVA), Q value = 1

Table S57. Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	34	90.9 (17.8)
subtype1	14	91.4 (8.6)
subtype2	12	87.5 (28.3)
subtype3	8	95.0 (7.6)

Figure S50. Get High-res Image Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'RNAseq CNMF subtypes' versus 'PATHOLOGY.T'

P value = 4.85e-06 (Chi-square test), Q value = 0.00031

Table S58. Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3	T4
ALL	238	60	171	11
subtype1	118	22	57	2
subtype2	59	27	88	6
subtype3	61	11	26	3

Figure S51. Get High-res Image Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

'RNAseq CNMF subtypes' versus 'PATHOLOGY.N'

P value = 0.0117 (Fisher's exact test), Q value = 0.43

Table S59. Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	228	17
subtype1	96	2
subtype2	85	12
subtype3	47	3

Figure S52. Get High-res Image Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

'RNAseq CNMF subtypes' versus 'PATHOLOGICSPREAD(M)'

P value = 0.000645 (Fisher's exact test), Q value = 0.033

Table S60. Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	403	77
subtype1	176	23
subtype2	136	44
subtype3	91	10

Figure S53. Get High-res Image Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'RNAseq CNMF subtypes' versus 'TUMOR.STAGE'

P value = 1.27e-06 (Chi-square test), Q value = 8.6e-05

Table S61. Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	234	48	117	81
subtype1	118	18	39	24
subtype2	56	21	58	45
subtype3	60	9	20	12

Figure S54. Get High-res Image Clustering Approach #7: 'RNAseq CNMF subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

Clustering Approach #8: 'RNAseq cHierClus subtypes'

Table S62. Get Full Table Description of clustering approach #8: 'RNAseq cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	73	215	192

'RNAseq cHierClus subtypes' versus 'Time to Death'

P value = 3.34e-08 (logrank test), Q value = 2.5e-06

Table S63. Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	478	155	0.1 - 111.0 (34.3)
subtype1	72	14	0.2 - 92.0 (27.8)
subtype2	214	48	0.1 - 111.0 (37.2)
subtype3	192	93	0.1 - 93.3 (30.5)

Figure S55. Get High-res Image Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq cHierClus subtypes' versus 'AGE'

P value = 0.358 (ANOVA), Q value = 1

Table S64. Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	479	60.6 (12.2)
subtype1	73	58.7 (13.2)
subtype2	214	61.0 (12.4)
subtype3	192	60.8 (11.5)

Figure S56. Get High-res Image Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'RNAseq cHierClus subtypes' versus 'GENDER'

P value = 0.00326 (Fisher's exact test), Q value = 0.15

Table S65. Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	167	313
subtype1	23	50
subtype2	92	123
subtype3	52	140

Figure S57. Get High-res Image Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

'RNAseq cHierClus subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.357 (ANOVA), Q value = 1

Table S66. Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	34	90.9 (17.8)
subtype1	10	95.0 (9.7)
subtype2	13	93.1 (6.3)
subtype3	11	84.5 (29.1)

Figure S58. Get High-res Image Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'RNAseq cHierClus subtypes' versus 'PATHOLOGY.T'

P value = 7.56e-12 (Chi-square test), Q value = 5.8e-10

Table S67. Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3	T4
ALL	238	60	171	11
subtype1	55	6	11	1
subtype2	126	27	61	1
subtype3	57	27	99	9

Figure S59. Get High-res Image Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

'RNAseq cHierClus subtypes' versus 'PATHOLOGY.N'

P value = 0.000966 (Fisher's exact test), Q value = 0.047

Table S68. Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	228	17
subtype1	32	2
subtype2	104	1
subtype3	92	14

Figure S60. Get High-res Image Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

'RNAseq cHierClus subtypes' versus 'PATHOLOGICSPREAD(M)'

P value = 2e-06 (Fisher's exact test), Q value = 0.00013

Table S69. Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	403	77
subtype1	70	3
subtype2	191	24
subtype3	142	50

Figure S61. Get High-res Image Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'RNAseq cHierClus subtypes' versus 'TUMOR.STAGE'

P value = 2.63e-12 (Chi-square test), Q value = 2.1e-10

Table S70. Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	234	48	117	81
subtype1	54	6	10	3
subtype2	126	21	43	25
subtype3	54	21	64	53

Figure S62. Get High-res Image Clustering Approach #8: 'RNAseq cHierClus subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

Clustering Approach #9: 'MIRseq CNMF subtypes'

Table S71. Get Full Table Description of clustering approach #9: 'MIRseq CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	209	106	165

'MIRseq CNMF subtypes' versus 'Time to Death'

P value = 8.12e-06 (logrank test), Q value = 5e-04

Table S72. Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	478	155	0.1 - 111.0 (35.2)
subtype1	209	47	0.1 - 111.0 (36.2)
subtype2	106	30	0.1 - 109.9 (35.8)
subtype3	163	78	0.2 - 93.3 (31.1)

Figure S63. Get High-res Image Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'MIRseq CNMF subtypes' versus 'AGE'

P value = 0.0537 (ANOVA), Q value = 1

Table S73. Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	480	60.6 (12.2)
subtype1	209	62.0 (12.2)
subtype2	106	58.7 (12.7)
subtype3	165	59.9 (11.7)

Figure S64. Get High-res Image Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #2: 'AGE'

'MIRseq CNMF subtypes' versus 'GENDER'

P value = 0.00265 (Fisher's exact test), Q value = 0.12

Table S74. Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	163	317
subtype1	88	121
subtype2	33	73
subtype3	42	123

Figure S65. Get High-res Image Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #3: 'GENDER'

'MIRseq CNMF subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.149 (ANOVA), Q value = 1

Table S75. Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	36	88.3 (23.0)
subtype1	16	91.9 (8.3)
subtype2	9	95.6 (7.3)
subtype3	11	77.3 (38.8)

Figure S66. Get High-res Image Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'MIRseq CNMF subtypes' versus 'PATHOLOGY.T'

P value = 0.00028 (Chi-square test), Q value = 0.015

Table S76. Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3	T4
ALL	233	62	174	11
subtype1	115	26	66	2
subtype2	62	8	32	4
subtype3	56	28	76	5

Figure S67. Get High-res Image Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

'MIRseq CNMF subtypes' versus 'PATHOLOGY.N'

P value = 0.00693 (Fisher's exact test), Q value = 0.3

Table S77. Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	222	17
subtype1	95	2
subtype2	51	3
subtype3	76	12

Figure S68. Get High-res Image Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

'MIRseq CNMF subtypes' versus 'PATHOLOGICSPREAD(M)'

P value = 0.000305 (Fisher's exact test), Q value = 0.016

Table S78. Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	405	75
subtype1	184	25
subtype2	97	9
subtype3	124	41

Figure S69. Get High-res Image Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'MIRseq CNMF subtypes' versus 'TUMOR.STAGE'

P value = 8.11e-05 (Chi-square test), Q value = 0.0047

Table S79. Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	229	50	122	79
subtype1	114	20	49	26
subtype2	60	8	28	10
subtype3	55	22	45	43

Figure S70. Get High-res Image Clustering Approach #9: 'MIRseq CNMF subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

Clustering Approach #10: 'MIRseq cHierClus subtypes'

Table S80. Get Full Table Description of clustering approach #10: 'MIRseq cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	51	281	148

'MIRseq cHierClus subtypes' versus 'Time to Death'

P value = 2.74e-05 (logrank test), Q value = 0.0016

Table S81. Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	478	155	0.1 - 111.0 (35.2)
subtype1	50	16	0.5 - 93.3 (39.6)
subtype2	280	70	0.1 - 111.0 (36.0)
subtype3	148	69	0.1 - 109.9 (29.5)

Figure S71. Get High-res Image Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'MIRseq cHierClus subtypes' versus 'AGE'

P value = 0.0521 (ANOVA), Q value = 1

Table S82. Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	480	60.6 (12.2)
subtype1	51	56.7 (12.1)
subtype2	281	61.1 (11.9)
subtype3	148	61.0 (12.6)

Figure S72. Get High-res Image Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'MIRseq cHierClus subtypes' versus 'GENDER'

P value = 0.182 (Fisher's exact test), Q value = 1

Table S83. Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	163	317
subtype1	15	36
subtype2	105	176
subtype3	43	105

Figure S73. Get High-res Image Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

'MIRseq cHierClus subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.849 (ANOVA), Q value = 1

Table S84. Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	36	88.3 (23.0)
subtype1	4	92.5 (9.6)
subtype2	19	86.3 (22.4)
subtype3	13	90.0 (27.4)

Figure S74. Get High-res Image Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'MIRseq cHierClus subtypes' versus 'PATHOLOGY.T'

P value = 0.00064 (Chi-square test), Q value = 0.033

Table S85. Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

nPatients	T1	T2	T3	T4
ALL	233	62	174	11
subtype1	22	12	15	2
subtype2	153	35	91	2
subtype3	58	15	68	7

Figure S75. Get High-res Image Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.T'

'MIRseq cHierClus subtypes' versus 'PATHOLOGY.N'

P value = 0.00892 (Fisher's exact test), Q value = 0.36

Table S86. Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

nPatients	0	1
ALL	222	17
subtype1	26	5
subtype2	128	4
subtype3	68	8

Figure S76. Get High-res Image Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.N'

'MIRseq cHierClus subtypes' versus 'PATHOLOGICSPREAD(M)'

P value = 0.282 (Fisher's exact test), Q value = 1

Table S87. Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

nPatients	M0	M1
ALL	405	75
subtype1	45	6
subtype2	241	40
subtype3	119	29

Figure S77. Get High-res Image Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #7: 'PATHOLOGICSPREAD(M)'

'MIRseq cHierClus subtypes' versus 'TUMOR.STAGE'

P value = 0.0143 (Chi-square test), Q value = 0.5

Table S88. Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

nPatients	I	II	III	IV
ALL	229	50	122	79
subtype1	22	10	13	6
subtype2	150	27	62	42
subtype3	57	13	47	31

Figure S78. Get High-res Image Clustering Approach #10: 'MIRseq cHierClus subtypes' versus Clinical Feature #8: 'TUMOR.STAGE'

Methods & Data

Input

Cluster data file = KIRC-TP.mergedcluster.txt
Clinical data file = KIRC-TP.clin.merged.picked.txt
Number of patients = 502
Number of clustering approaches = 10
Number of selected clinical features = 8
Exclude small clusters that include fewer than K patients, K = 3

Clustering approaches

CNMF clustering

consensus non-negative matrix factorization clustering approach (Brunet et al. 2004)

Consensus hierarchical clustering

Resampling-based clustering method (Monti et al. 2003)

Survival analysis

For survival clinical features, the Kaplan-Meier survival curves of tumors with and without gene mutations were plotted and the statistical significance P values were estimated by logrank test (Bland and Altman 2004) using the 'survdiff' function in R

ANOVA analysis

For continuous numerical clinical features, one-way analysis of variance (Howell 2002) was applied to compare the clinical values between tumor subtypes using 'anova' function in R

Fisher's exact test

For binary clinical features, two-tailed Fisher's exact tests (Fisher 1922) were used to estimate the P values using the 'fisher.test' function in R

Chi-square test

For multi-class clinical features (nominal or ordinal), Chi-square tests (Greenwood and Nikulin 1996) were used to estimate the P values using the 'chisq.test' function in R

Q value calculation

For multiple hypothesis correction, Q value is the False Discovery Rate (FDR) analogue of the P value (Benjamini and Hochberg 1995), defined as the minimum FDR at which the test may be called significant. We used the 'Benjamini and Hochberg' method of 'p.adjust' function in R to convert P values into Q values.

References

[1] Brunet et al., Metagenes and molecular pattern discovery using matrix factorization, PNAS 101(12):4164-9 (2004)

[2] Monti et al., Consensus Clustering: A Resampling-Based Method for Class Discovery and Visualization of Gene Expression Microarray Data, Machine Learning 52(1):91-118 (2003)

[3] Bland and Altman, Statistics notes: The logrank test, BMJ 328(7447):1073 (2004)

[4] Howell, D, Statistical Methods for Psychology. (5th ed.), Duxbury Press:324-5 (2002)

[5] Fisher, R.A., On the interpretation of chi-square from contingency tables, and the calculation of P, Journal of the Royal Statistical Society 85(1):87-94 (1922)

[6] Greenwood and Nikulin, A guide to chi-squared testing, Wiley, New York. ISBN 047155779X (1996)

[7] Benjamini and Hochberg, Controlling the false discovery rate: a practical and powerful approach to multiple testing, Journal of the Royal Statistical Society Series B 59:289-300 (1995)

Made with Nozzle