Correlation between aggregated molecular cancer subtypes and selected clinical features

Prostate Adenocarcinoma (Primary solid tumor)

21 August 2015 | analyses__2015_08_21

Maintainer Information

Citation Information

Maintained by TCGA GDAC Team (Broad Institute/MD Anderson Cancer Center/Harvard Medical School)

Cite as Broad Institute TCGA Genome Data Analysis Center (2015): Correlation between aggregated molecular cancer subtypes and selected clinical features. Broad Institute of MIT and Harvard. doi:10.7908/C1MP52JF

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 11 clinical features across 497 patients, 54 significant findings detected with P value < 0.05 and Q value < 0.25.

3 subtypes identified in current cancer cohort by 'Copy Number Ratio CNMF subtypes'. These subtypes correlate to 'YEARS_TO_BIRTH', 'PATHOLOGY_T_STAGE', 'PATHOLOGY_N_STAGE', 'RADIATION_THERAPY', 'RESIDUAL_TUMOR', 'NUMBER_OF_LYMPH_NODES', 'GLEASON_SCORE', and 'PSA_VALUE'.
3 subtypes identified in current cancer cohort by 'METHLYATION CNMF'. These subtypes correlate to 'YEARS_TO_BIRTH', 'PATHOLOGY_T_STAGE', 'RESIDUAL_TUMOR', and 'GLEASON_SCORE'.
CNMF clustering analysis on RPPA data identified 3 subtypes that correlate to 'PATHOLOGY_T_STAGE', 'RESIDUAL_TUMOR', 'NUMBER_OF_LYMPH_NODES', 'GLEASON_SCORE', 'PSA_VALUE', and 'RACE'.
Consensus hierarchical clustering analysis on RPPA data identified 4 subtypes that correlate to 'YEARS_TO_BIRTH', 'PATHOLOGY_T_STAGE', 'RESIDUAL_TUMOR', and 'GLEASON_SCORE'.
CNMF clustering analysis on sequencing-based mRNA expression data identified 3 subtypes that correlate to 'PATHOLOGY_T_STAGE', 'NUMBER_OF_LYMPH_NODES', 'GLEASON_SCORE', and 'PSA_VALUE'.
Consensus hierarchical clustering analysis on sequencing-based mRNA expression data identified 3 subtypes that correlate to 'YEARS_TO_BIRTH', 'PATHOLOGY_T_STAGE', 'PATHOLOGY_N_STAGE', 'RESIDUAL_TUMOR', 'NUMBER_OF_LYMPH_NODES', 'GLEASON_SCORE', and 'RACE'.
3 subtypes identified in current cancer cohort by 'MIRSEQ CNMF'. These subtypes correlate to 'PATHOLOGY_T_STAGE', 'PATHOLOGY_N_STAGE', 'NUMBER_OF_LYMPH_NODES', 'GLEASON_SCORE', and 'PSA_VALUE'.
5 subtypes identified in current cancer cohort by 'MIRSEQ CHIERARCHICAL'. These subtypes correlate to 'YEARS_TO_BIRTH', 'PATHOLOGY_T_STAGE', 'PATHOLOGY_N_STAGE', 'RESIDUAL_TUMOR', 'NUMBER_OF_LYMPH_NODES', and 'GLEASON_SCORE'.
4 subtypes identified in current cancer cohort by 'MIRseq Mature CNMF subtypes'. These subtypes correlate to 'YEARS_TO_BIRTH', 'PATHOLOGY_T_STAGE', 'RESIDUAL_TUMOR', and 'GLEASON_SCORE'.
6 subtypes identified in current cancer cohort by 'MIRseq Mature cHierClus subtypes'. These subtypes correlate to 'YEARS_TO_BIRTH', 'PATHOLOGY_T_STAGE', 'PATHOLOGY_N_STAGE', 'RESIDUAL_TUMOR', 'NUMBER_OF_LYMPH_NODES', and 'GLEASON_SCORE'.

Results

Overview of the results

Table 1. Get Full Table Overview of the association between subtypes identified by 10 different clustering approaches and 11 clinical features. Shown in the table are P values (Q values). Thresholded by P value < 0.05 and Q value < 0.25, 54 significant findings detected.


Clinical Features	Statistical Tests	Copy Number Ratio CNMF subtypes	METHLYATION CNMF	RPPA CNMF subtypes	RPPA cHierClus subtypes	RNAseq CNMF subtypes	RNAseq cHierClus subtypes	MIRSEQ CNMF	MIRSEQ CHIERARCHICAL	MIRseq Mature CNMF subtypes	MIRseq Mature cHierClus subtypes
Time to Death	logrank test	0.753 (0.804)	0.471 (0.589)	0.728 (0.789)	0.662 (0.769)	0.731 (0.789)	0.914 (0.94)	0.425 (0.537)	0.975 (0.993)	0.68 (0.771)	0.985 (0.994)
YEARS TO BIRTH	Kruskal-Wallis (anova)	0.00239 (0.00822)	0.00102 (0.00402)	0.137 (0.221)	0.00234 (0.00822)	0.155 (0.245)	0.012 (0.0339)	0.166 (0.257)	0.0247 (0.059)	0.0061 (0.0186)	0.0386 (0.0832)
PATHOLOGY T STAGE	Fisher's exact test	1e-05 (7.86e-05)	7e-05 (0.000405)	0.00019 (0.00095)	0.00388 (0.0129)	0.00016 (0.000838)	0.00196 (0.00719)	0.00826 (0.0246)	1e-05 (7.86e-05)	0.00553 (0.0179)	2e-05 (0.000137)
PATHOLOGY N STAGE	Fisher's exact test	1e-05 (7.86e-05)	0.365 (0.483)	0.0507 (0.101)	0.0907 (0.158)	0.0881 (0.156)	0.0477 (0.0991)	0.0116 (0.0335)	0.00048 (0.00203)	0.214 (0.318)	2e-05 (0.000137)
RADIATION THERAPY	Fisher's exact test	0.0013 (0.00493)	0.396 (0.512)	0.176 (0.268)	0.109 (0.182)	0.542 (0.669)	0.0728 (0.134)	0.259 (0.36)	0.0677 (0.128)	0.423 (0.537)	0.339 (0.455)
HISTOLOGICAL TYPE	Fisher's exact test	0.206 (0.311)	0.678 (0.771)	0.732 (0.789)	0.0864 (0.156)	0.369 (0.483)	0.243 (0.35)	0.109 (0.182)	0.799 (0.834)	0.156 (0.245)	0.664 (0.769)
RESIDUAL TUMOR	Fisher's exact test	1e-05 (7.86e-05)	0.0417 (0.0883)	0.00058 (0.00236)	0.0186 (0.0506)	0.0731 (0.134)	0.0193 (0.0506)	0.127 (0.209)	6e-05 (0.000367)	0.00034 (0.0015)	0.00012 (0.00066)
NUMBER OF LYMPH NODES	Kruskal-Wallis (anova)	1.72e-08 (4.72e-07)	0.271 (0.372)	0.0223 (0.0558)	0.0931 (0.16)	0.0382 (0.0832)	0.0315 (0.0722)	0.006 (0.0186)	5.3e-05 (0.000343)	0.257 (0.36)	1.54e-06 (2.42e-05)
GLEASON SCORE	Kruskal-Wallis (anova)	2.94e-23 (3.23e-21)	0.000326 (0.0015)	1.48e-06 (2.42e-05)	3.2e-06 (3.52e-05)	0.000258 (0.00123)	2.95e-06 (3.52e-05)	2.48e-06 (3.41e-05)	9.1e-12 (5.01e-10)	4.2e-07 (9.23e-06)	1.98e-11 (7.26e-10)
PSA VALUE	Kruskal-Wallis (anova)	0.0228 (0.0558)	0.605 (0.717)	0.0192 (0.0506)	0.0536 (0.105)	0.0498 (0.101)	0.607 (0.717)	0.0359 (0.0805)	0.717 (0.789)	0.0559 (0.108)	0.56 (0.679)
RACE	Fisher's exact test	0.562 (0.679)	1 (1.00)	0.0305 (0.0715)	0.245 (0.35)	0.336 (0.455)	0.0228 (0.0558)	0.217 (0.318)	0.708 (0.789)	0.802 (0.834)	0.804 (0.834)

Clustering Approach #1: 'Copy Number Ratio CNMF subtypes'

Table S1. Description of clustering approach #1: 'Copy Number Ratio CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	273	133	85

'Copy Number Ratio CNMF subtypes' versus 'Time to Death'

P value = 0.753 (logrank test), Q value = 0.8

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

	nPatients	nDeath	Duration Range (Median), Month
ALL	489	9	0.7 - 165.2 (28.8)
subtype1	272	4	0.7 - 165.2 (28.0)
subtype2	132	4	0.8 - 141.2 (32.2)
subtype3	85	1	1.0 - 102.9 (28.8)

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

'Copy Number Ratio CNMF subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.00239 (Kruskal-Wallis (anova)), Q value = 0.0082

Table S3. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	480	61.0 (6.8)
subtype1	267	60.2 (6.8)
subtype2	130	62.7 (6.2)
subtype3	83	60.7 (7.4)

Figure S2. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'Copy Number Ratio CNMF subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 1e-05 (Fisher's exact test), Q value = 7.9e-05

Table S4. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	188	287	10
subtype1	129	139	2
subtype2	15	109	7
subtype3	44	39	1

Figure S3. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'Copy Number Ratio CNMF subtypes' versus 'PATHOLOGY_N_STAGE'

P value = 1e-05 (Fisher's exact test), Q value = 7.9e-05

Table S5. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	342	77
subtype1	199	26
subtype2	82	43
subtype3	61	8

Figure S4. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'Copy Number Ratio CNMF subtypes' versus 'RADIATION_THERAPY'

P value = 0.0013 (Fisher's exact test), Q value = 0.0049

Table S6. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	378	57
subtype1	220	22
subtype2	90	27
subtype3	68	8

Figure S5. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

'Copy Number Ratio CNMF subtypes' versus 'HISTOLOGICAL_TYPE'

P value = 0.206 (Fisher's exact test), Q value = 0.31

Table S7. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	14	477
subtype1	9	264
subtype2	5	128
subtype3	0	85

Figure S6. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'Copy Number Ratio CNMF subtypes' versus 'RESIDUAL_TUMOR'

P value = 1e-05 (Fisher's exact test), Q value = 7.9e-05

Table S8. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	312	143	5	15
subtype1	199	55	3	7
subtype2	61	64	1	4
subtype3	52	24	1	4

Figure S7. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'Copy Number Ratio CNMF subtypes' versus 'NUMBER_OF_LYMPH_NODES'

P value = 1.72e-08 (Kruskal-Wallis (anova)), Q value = 4.7e-07

Table S9. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	401	0.4 (1.4)
subtype1	215	0.2 (1.1)
subtype2	118	1.0 (1.9)
subtype3	68	0.1 (0.3)

Figure S8. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'Copy Number Ratio CNMF subtypes' versus 'GLEASON_SCORE'

P value = 2.94e-23 (Kruskal-Wallis (anova)), Q value = 3.2e-21

Table S10. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	491	7.6 (1.0)
subtype1	273	7.3 (0.9)
subtype2	133	8.4 (0.9)
subtype3	85	7.4 (0.9)

Figure S9. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

'Copy Number Ratio CNMF subtypes' versus 'PSA_VALUE'

P value = 0.0228 (Kruskal-Wallis (anova)), Q value = 0.056

Table S11. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	434	1.8 (16.0)
subtype1	245	0.7 (3.0)
subtype2	114	4.9 (30.7)
subtype3	75	0.6 (2.2)

Figure S10. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #10: 'PSA_VALUE'

'Copy Number Ratio CNMF subtypes' versus 'RACE'

P value = 0.562 (Fisher's exact test), Q value = 0.68

Table S12. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	2	7	146
subtype1	2	3	90
subtype2	0	1	29
subtype3	0	3	27

Figure S11. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #11: 'RACE'

Clustering Approach #2: 'METHLYATION CNMF'

Table S13. Description of clustering approach #2: 'METHLYATION CNMF'

Cluster Labels	1	2	3
Number of samples	160	168	169

'METHLYATION CNMF' versus 'Time to Death'

P value = 0.471 (logrank test), Q value = 0.59

Table S14. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	495	10	0.7 - 165.2 (28.8)
subtype1	160	1	1.0 - 102.9 (27.8)
subtype2	167	5	0.7 - 122.2 (31.7)
subtype3	168	4	1.0 - 165.2 (28.4)

Figure S12. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #1: 'Time to Death'

'METHLYATION CNMF' versus 'YEARS_TO_BIRTH'

P value = 0.00102 (Kruskal-Wallis (anova)), Q value = 0.004

Table S15. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	486	61.0 (6.8)
subtype1	156	59.8 (6.9)
subtype2	164	62.5 (6.9)
subtype3	166	60.6 (6.4)

Figure S13. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'METHLYATION CNMF' versus 'PATHOLOGY_T_STAGE'

P value = 7e-05 (Fisher's exact test), Q value = 0.00041

Table S16. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	189	292	10
subtype1	81	76	0
subtype2	57	101	7
subtype3	51	115	3

Figure S14. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'METHLYATION CNMF' versus 'PATHOLOGY_N_STAGE'

P value = 0.365 (Fisher's exact test), Q value = 0.48

Table S17. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	345	79
subtype1	114	21
subtype2	118	26
subtype3	113	32

Figure S15. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'METHLYATION CNMF' versus 'RADIATION_THERAPY'

P value = 0.396 (Fisher's exact test), Q value = 0.51

Table S18. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	383	57
subtype1	123	22
subtype2	134	15
subtype3	126	20

Figure S16. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #5: 'RADIATION_THERAPY'

'METHLYATION CNMF' versus 'HISTOLOGICAL_TYPE'

P value = 0.678 (Fisher's exact test), Q value = 0.77

Table S19. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	15	482
subtype1	4	156
subtype2	4	164
subtype3	7	162

Figure S17. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'METHLYATION CNMF' versus 'RESIDUAL_TUMOR'

P value = 0.0417 (Fisher's exact test), Q value = 0.088

Table S20. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	315	146	5	15
subtype1	113	37	1	4
subtype2	91	64	1	5
subtype3	111	45	3	6

Figure S18. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'METHLYATION CNMF' versus 'NUMBER_OF_LYMPH_NODES'

P value = 0.271 (Kruskal-Wallis (anova)), Q value = 0.37

Table S21. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	406	0.4 (1.4)
subtype1	128	0.3 (0.8)
subtype2	138	0.4 (1.4)
subtype3	140	0.6 (1.7)

Figure S19. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'METHLYATION CNMF' versus 'GLEASON_SCORE'

P value = 0.000326 (Kruskal-Wallis (anova)), Q value = 0.0015

Table S22. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	497	7.6 (1.0)
subtype1	160	7.4 (1.0)
subtype2	168	7.8 (1.0)
subtype3	169	7.6 (1.0)

Figure S20. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #9: 'GLEASON_SCORE'

'METHLYATION CNMF' versus 'PSA_VALUE'

P value = 0.605 (Kruskal-Wallis (anova)), Q value = 0.72

Table S23. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	440	1.8 (15.9)
subtype1	145	0.7 (3.2)
subtype2	142	3.4 (27.3)
subtype3	153	1.3 (5.0)

Figure S21. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #10: 'PSA_VALUE'

'METHLYATION CNMF' versus 'RACE'

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

Table S24. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	2	7	147
subtype1	1	2	53
subtype2	0	2	37
subtype3	1	3	57

Figure S22. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #11: 'RACE'

Clustering Approach #3: 'RPPA CNMF subtypes'

Table S25. Description of clustering approach #3: 'RPPA CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	124	69	159

'RPPA CNMF subtypes' versus 'Time to Death'

P value = 0.728 (logrank test), Q value = 0.79

Table S26. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	351	7	0.7 - 165.2 (31.0)
subtype1	124	2	0.9 - 140.2 (32.2)
subtype2	69	1	0.7 - 115.1 (25.9)
subtype3	158	4	0.8 - 165.2 (31.2)

Figure S23. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'RPPA CNMF subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.137 (Kruskal-Wallis (anova)), Q value = 0.22

Table S27. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	349	61.4 (6.6)
subtype1	122	61.0 (7.3)
subtype2	69	62.7 (6.1)
subtype3	158	61.2 (6.2)

Figure S24. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'RPPA CNMF subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.00019 (Fisher's exact test), Q value = 0.00095

Table S28. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	111	227	10
subtype1	52	68	2
subtype2	28	39	1
subtype3	31	120	7

Figure S25. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'RPPA CNMF subtypes' versus 'PATHOLOGY_N_STAGE'

P value = 0.0507 (Fisher's exact test), Q value = 0.1

Table S29. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	246	63
subtype1	89	15
subtype2	55	11
subtype3	102	37

Figure S26. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'RPPA CNMF subtypes' versus 'RADIATION_THERAPY'

P value = 0.176 (Fisher's exact test), Q value = 0.27

Table S30. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	278	40
subtype1	107	12
subtype2	55	5
subtype3	116	23

Figure S27. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

'RPPA CNMF subtypes' versus 'HISTOLOGICAL_TYPE'

P value = 0.732 (Fisher's exact test), Q value = 0.79

Table S31. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	11	341
subtype1	3	121
subtype2	3	66
subtype3	5	154

Figure S28. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'RPPA CNMF subtypes' versus 'RESIDUAL_TUMOR'

P value = 0.00058 (Fisher's exact test), Q value = 0.0024

Table S32. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	220	108	3	10
subtype1	82	33	1	2
subtype2	56	12	0	0
subtype3	82	63	2	8

Figure S29. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'RPPA CNMF subtypes' versus 'NUMBER_OF_LYMPH_NODES'

P value = 0.0223 (Kruskal-Wallis (anova)), Q value = 0.056

Table S33. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	300	0.5 (1.5)
subtype1	102	0.3 (1.0)
subtype2	66	0.3 (0.7)
subtype3	132	0.8 (2.1)

Figure S30. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'RPPA CNMF subtypes' versus 'GLEASON_SCORE'

P value = 1.48e-06 (Kruskal-Wallis (anova)), Q value = 2.4e-05

Table S34. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	352	7.6 (1.0)
subtype1	124	7.4 (1.0)
subtype2	69	7.4 (0.9)
subtype3	159	7.9 (1.0)

Figure S31. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

'RPPA CNMF subtypes' versus 'PSA_VALUE'

P value = 0.0192 (Kruskal-Wallis (anova)), Q value = 0.051

Table S35. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	318	1.3 (4.7)
subtype1	116	0.7 (2.9)
subtype2	65	1.2 (4.5)
subtype3	137	1.9 (5.8)

Figure S32. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #10: 'PSA_VALUE'

'RPPA CNMF subtypes' versus 'RACE'

P value = 0.0305 (Fisher's exact test), Q value = 0.071

Table S36. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	2	6	121
subtype1	2	1	41
subtype2	0	0	38
subtype3	0	5	42

Figure S33. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #11: 'RACE'

Clustering Approach #4: 'RPPA cHierClus subtypes'

Table S37. Description of clustering approach #4: 'RPPA cHierClus subtypes'

Cluster Labels	1	2	3	4
Number of samples	75	114	93	70

'RPPA cHierClus subtypes' versus 'Time to Death'

P value = 0.662 (logrank test), Q value = 0.77

Table S38. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	351	7	0.7 - 165.2 (31.0)
subtype1	75	1	0.9 - 109.2 (38.0)
subtype2	113	3	0.7 - 140.2 (23.3)
subtype3	93	1	1.6 - 165.2 (33.1)
subtype4	70	2	0.9 - 113.1 (31.2)

Figure S34. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'RPPA cHierClus subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.00234 (Kruskal-Wallis (anova)), Q value = 0.0082

Table S39. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	349	61.4 (6.6)
subtype1	75	60.6 (7.2)
subtype2	112	62.8 (6.4)
subtype3	92	59.7 (6.5)
subtype4	70	62.5 (5.8)

Figure S35. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'RPPA cHierClus subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.00388 (Fisher's exact test), Q value = 0.013

Table S40. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	111	227	10
subtype1	34	38	1
subtype2	39	73	1
subtype3	24	66	3
subtype4	14	50	5

Figure S36. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'RPPA cHierClus subtypes' versus 'PATHOLOGY_N_STAGE'

P value = 0.0907 (Fisher's exact test), Q value = 0.16

Table S41. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	246	63
subtype1	55	6
subtype2	79	26
subtype3	62	15
subtype4	50	16

Figure S37. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'RPPA cHierClus subtypes' versus 'RADIATION_THERAPY'

P value = 0.109 (Fisher's exact test), Q value = 0.18

Table S42. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	278	40
subtype1	70	5
subtype2	80	16
subtype3	76	8
subtype4	52	11

Figure S38. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

'RPPA cHierClus subtypes' versus 'HISTOLOGICAL_TYPE'

P value = 0.0864 (Fisher's exact test), Q value = 0.16

Table S43. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	11	341
subtype1	2	73
subtype2	5	109
subtype3	0	93
subtype4	4	66

Figure S39. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'RPPA cHierClus subtypes' versus 'RESIDUAL_TUMOR'

P value = 0.0186 (Fisher's exact test), Q value = 0.051

Table S44. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	220	108	3	10
subtype1	56	15	0	0
subtype2	74	32	2	4
subtype3	56	29	1	3
subtype4	34	32	0	3

Figure S40. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'RPPA cHierClus subtypes' versus 'NUMBER_OF_LYMPH_NODES'

P value = 0.0931 (Kruskal-Wallis (anova)), Q value = 0.16

Table S45. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	300	0.5 (1.5)
subtype1	60	0.2 (0.8)
subtype2	103	0.5 (1.2)
subtype3	75	0.5 (2.0)
subtype4	62	0.8 (1.9)

Figure S41. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'RPPA cHierClus subtypes' versus 'GLEASON_SCORE'

P value = 3.2e-06 (Kruskal-Wallis (anova)), Q value = 3.5e-05

Table S46. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	352	7.6 (1.0)
subtype1	75	7.3 (0.9)
subtype2	114	7.6 (1.0)
subtype3	93	7.6 (0.9)
subtype4	70	8.2 (1.0)

Figure S42. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

'RPPA cHierClus subtypes' versus 'PSA_VALUE'

P value = 0.0536 (Kruskal-Wallis (anova)), Q value = 0.11

Table S47. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	318	1.3 (4.7)
subtype1	72	0.4 (2.3)
subtype2	105	1.8 (5.8)
subtype3	81	0.6 (1.7)
subtype4	60	2.5 (6.6)

Figure S43. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #10: 'PSA_VALUE'

'RPPA cHierClus subtypes' versus 'RACE'

P value = 0.245 (Fisher's exact test), Q value = 0.35

Table S48. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	2	6	121
subtype1	1	1	27
subtype2	0	1	47
subtype3	1	1	27
subtype4	0	3	20

Figure S44. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #11: 'RACE'

Clustering Approach #5: 'RNAseq CNMF subtypes'

Table S49. Description of clustering approach #5: 'RNAseq CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	147	175	174

'RNAseq CNMF subtypes' versus 'Time to Death'

P value = 0.731 (logrank test), Q value = 0.79

Table S50. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	494	10	0.7 - 165.2 (28.8)
subtype1	146	3	0.7 - 102.9 (27.5)
subtype2	175	3	0.8 - 122.2 (31.7)
subtype3	173	4	0.9 - 165.2 (28.2)

Figure S45. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq CNMF subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.155 (Kruskal-Wallis (anova)), Q value = 0.24

Table S51. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	485	61.0 (6.8)
subtype1	143	60.3 (7.2)
subtype2	170	61.8 (6.9)
subtype3	172	60.9 (6.3)

Figure S46. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'RNAseq CNMF subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.00016 (Fisher's exact test), Q value = 0.00084

Table S52. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	188	292	10
subtype1	73	70	1
subtype2	70	99	4
subtype3	45	123	5

Figure S47. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'RNAseq CNMF subtypes' versus 'PATHOLOGY_N_STAGE'

P value = 0.0881 (Fisher's exact test), Q value = 0.16

Table S53. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	344	79
subtype1	106	17
subtype2	121	25
subtype3	117	37

Figure S48. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'RNAseq CNMF subtypes' versus 'RADIATION_THERAPY'

P value = 0.542 (Fisher's exact test), Q value = 0.67

Table S54. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	382	57
subtype1	113	18
subtype2	141	17
subtype3	128	22

Figure S49. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

'RNAseq CNMF subtypes' versus 'HISTOLOGICAL_TYPE'

P value = 0.369 (Fisher's exact test), Q value = 0.48

Table S55. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	15	481
subtype1	3	144
subtype2	4	171
subtype3	8	166

Figure S50. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'RNAseq CNMF subtypes' versus 'RESIDUAL_TUMOR'

P value = 0.0731 (Fisher's exact test), Q value = 0.13

Table S56. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	314	146	5	15
subtype1	103	32	0	5
subtype2	101	64	2	4
subtype3	110	50	3	6

Figure S51. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'RNAseq CNMF subtypes' versus 'NUMBER_OF_LYMPH_NODES'

P value = 0.0382 (Kruskal-Wallis (anova)), Q value = 0.083

Table S57. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	405	0.4 (1.4)
subtype1	118	0.2 (0.6)
subtype2	141	0.4 (1.1)
subtype3	146	0.7 (1.9)

Figure S52. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'RNAseq CNMF subtypes' versus 'GLEASON_SCORE'

P value = 0.000258 (Kruskal-Wallis (anova)), Q value = 0.0012

Table S58. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	496	7.6 (1.0)
subtype1	147	7.3 (0.8)
subtype2	175	7.7 (1.1)
subtype3	174	7.8 (1.0)

Figure S53. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

'RNAseq CNMF subtypes' versus 'PSA_VALUE'

P value = 0.0498 (Kruskal-Wallis (anova)), Q value = 0.1

Table S59. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	439	1.8 (15.9)
subtype1	134	0.7 (3.3)
subtype2	149	1.0 (3.6)
subtype3	156	3.5 (26.2)

Figure S54. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #10: 'PSA_VALUE'

'RNAseq CNMF subtypes' versus 'RACE'

P value = 0.336 (Fisher's exact test), Q value = 0.45

Table S60. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	2	7	147
subtype1	2	1	51
subtype2	0	3	40
subtype3	0	3	56

Figure S55. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #11: 'RACE'

Clustering Approach #6: 'RNAseq cHierClus subtypes'

Table S61. Description of clustering approach #6: 'RNAseq cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	173	207	116

'RNAseq cHierClus subtypes' versus 'Time to Death'

P value = 0.914 (logrank test), Q value = 0.94

Table S62. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	494	10	0.7 - 165.2 (28.8)
subtype1	172	3	0.7 - 122.2 (27.8)
subtype2	207	4	0.8 - 114.4 (29.9)
subtype3	115	3	1.0 - 165.2 (27.8)

Figure S56. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq cHierClus subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.012 (Kruskal-Wallis (anova)), Q value = 0.034

Table S63. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	485	61.0 (6.8)
subtype1	168	60.1 (6.8)
subtype2	203	62.2 (6.9)
subtype3	114	60.3 (6.4)

Figure S57. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'RNAseq cHierClus subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.00196 (Fisher's exact test), Q value = 0.0072

Table S64. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	188	292	10
subtype1	85	84	2
subtype2	70	127	6
subtype3	33	81	2

Figure S58. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'RNAseq cHierClus subtypes' versus 'PATHOLOGY_N_STAGE'

P value = 0.0477 (Fisher's exact test), Q value = 0.099

Table S65. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	344	79
subtype1	127	18
subtype2	138	37
subtype3	79	24

Figure S59. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'RNAseq cHierClus subtypes' versus 'RADIATION_THERAPY'

P value = 0.0728 (Fisher's exact test), Q value = 0.13

Table S66. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	382	57
subtype1	140	17
subtype2	162	20
subtype3	80	20

Figure S60. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

'RNAseq cHierClus subtypes' versus 'HISTOLOGICAL_TYPE'

P value = 0.243 (Fisher's exact test), Q value = 0.35

Table S67. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	15	481
subtype1	3	170
subtype2	6	201
subtype3	6	110

Figure S61. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'RNAseq cHierClus subtypes' versus 'RESIDUAL_TUMOR'

P value = 0.0193 (Fisher's exact test), Q value = 0.051

Table S68. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	314	146	5	15
subtype1	121	38	0	6
subtype2	125	71	3	3
subtype3	68	37	2	6

Figure S62. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'RNAseq cHierClus subtypes' versus 'NUMBER_OF_LYMPH_NODES'

P value = 0.0315 (Kruskal-Wallis (anova)), Q value = 0.072

Table S69. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	405	0.4 (1.4)
subtype1	137	0.2 (0.5)
subtype2	170	0.5 (1.4)
subtype3	98	0.7 (1.9)

Figure S63. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'RNAseq cHierClus subtypes' versus 'GLEASON_SCORE'

P value = 2.95e-06 (Kruskal-Wallis (anova)), Q value = 3.5e-05

Table S70. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	496	7.6 (1.0)
subtype1	173	7.3 (0.9)
subtype2	207	7.8 (1.1)
subtype3	116	7.7 (1.0)

Figure S64. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

'RNAseq cHierClus subtypes' versus 'PSA_VALUE'

P value = 0.607 (Kruskal-Wallis (anova)), Q value = 0.72

Table S71. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	439	1.8 (15.9)
subtype1	156	0.8 (3.2)
subtype2	179	2.8 (24.3)
subtype3	104	1.7 (5.9)

Figure S65. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #10: 'PSA_VALUE'

'RNAseq cHierClus subtypes' versus 'RACE'

P value = 0.0228 (Fisher's exact test), Q value = 0.056

Table S72. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	2	7	147
subtype1	2	0	65
subtype2	0	5	46
subtype3	0	2	36

Figure S66. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #11: 'RACE'

Clustering Approach #7: 'MIRSEQ CNMF'

Table S73. Description of clustering approach #7: 'MIRSEQ CNMF'

Cluster Labels	1	2	3
Number of samples	159	147	187

'MIRSEQ CNMF' versus 'Time to Death'

P value = 0.425 (logrank test), Q value = 0.54

Table S74. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	491	10	0.7 - 165.2 (28.7)
subtype1	159	3	0.9 - 141.2 (25.1)
subtype2	147	1	0.7 - 115.1 (30.8)
subtype3	185	6	0.8 - 165.2 (30.6)

Figure S67. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #1: 'Time to Death'

'MIRSEQ CNMF' versus 'YEARS_TO_BIRTH'

P value = 0.166 (Kruskal-Wallis (anova)), Q value = 0.26

Table S75. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	482	61.0 (6.8)
subtype1	156	60.3 (6.6)
subtype2	144	60.7 (7.2)
subtype3	182	61.8 (6.6)

Figure S68. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'MIRSEQ CNMF' versus 'PATHOLOGY_T_STAGE'

P value = 0.00826 (Fisher's exact test), Q value = 0.025

Table S76. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	188	290	9
subtype1	64	90	2
subtype2	69	73	4
subtype3	55	127	3

Figure S69. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'MIRSEQ CNMF' versus 'PATHOLOGY_N_STAGE'

P value = 0.0116 (Fisher's exact test), Q value = 0.034

Table S77. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	343	78
subtype1	111	26
subtype2	105	12
subtype3	127	40

Figure S70. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'MIRSEQ CNMF' versus 'RADIATION_THERAPY'

P value = 0.259 (Fisher's exact test), Q value = 0.36

Table S78. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	380	57
subtype1	121	21
subtype2	121	12
subtype3	138	24

Figure S71. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #5: 'RADIATION_THERAPY'

'MIRSEQ CNMF' versus 'HISTOLOGICAL_TYPE'

P value = 0.109 (Fisher's exact test), Q value = 0.18

Table S79. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	15	478
subtype1	7	152
subtype2	1	146
subtype3	7	180

Figure S72. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'MIRSEQ CNMF' versus 'RESIDUAL_TUMOR'

P value = 0.127 (Fisher's exact test), Q value = 0.21

Table S80. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	312	145	5	15
subtype1	109	38	0	6
subtype2	93	43	3	2
subtype3	110	64	2	7

Figure S73. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'MIRSEQ CNMF' versus 'NUMBER_OF_LYMPH_NODES'

P value = 0.006 (Kruskal-Wallis (anova)), Q value = 0.019

Table S81. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	403	0.4 (1.4)
subtype1	127	0.3 (1.1)
subtype2	115	0.2 (0.7)
subtype3	161	0.7 (1.8)

Figure S74. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'MIRSEQ CNMF' versus 'GLEASON_SCORE'

P value = 2.48e-06 (Kruskal-Wallis (anova)), Q value = 3.4e-05

Table S82. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	493	7.6 (1.0)
subtype1	159	7.5 (0.9)
subtype2	147	7.4 (1.0)
subtype3	187	7.9 (1.0)

Figure S75. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #9: 'GLEASON_SCORE'

'MIRSEQ CNMF' versus 'PSA_VALUE'

P value = 0.0359 (Kruskal-Wallis (anova)), Q value = 0.081

Table S83. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	438	1.8 (15.9)
subtype1	146	1.0 (3.9)
subtype2	129	0.7 (3.9)
subtype3	163	3.4 (25.6)

Figure S76. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #10: 'PSA_VALUE'

'MIRSEQ CNMF' versus 'RACE'

P value = 0.217 (Fisher's exact test), Q value = 0.32

Table S84. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	2	7	146
subtype1	1	0	39
subtype2	1	3	69
subtype3	0	4	38

Figure S77. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #11: 'RACE'

Clustering Approach #8: 'MIRSEQ CHIERARCHICAL'

Table S85. Description of clustering approach #8: 'MIRSEQ CHIERARCHICAL'

Cluster Labels	1	2	3	4	5
Number of samples	148	98	59	132	56

'MIRSEQ CHIERARCHICAL' versus 'Time to Death'

P value = 0.975 (logrank test), Q value = 0.99

Table S86. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	491	10	0.7 - 165.2 (28.7)
subtype1	148	2	1.0 - 141.2 (27.6)
subtype2	98	2	3.7 - 115.1 (36.2)
subtype3	59	2	1.0 - 140.2 (24.3)
subtype4	131	3	0.8 - 165.2 (29.2)
subtype5	55	1	0.7 - 119.4 (28.3)

Figure S78. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #1: 'Time to Death'

'MIRSEQ CHIERARCHICAL' versus 'YEARS_TO_BIRTH'

P value = 0.0247 (Kruskal-Wallis (anova)), Q value = 0.059

Table S87. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	482	61.0 (6.8)
subtype1	145	60.5 (6.6)
subtype2	95	60.7 (6.7)
subtype3	59	59.6 (6.4)
subtype4	129	62.7 (6.7)
subtype5	54	60.4 (7.6)

Figure S79. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'MIRSEQ CHIERARCHICAL' versus 'PATHOLOGY_T_STAGE'

P value = 1e-05 (Fisher's exact test), Q value = 7.9e-05

Table S88. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	188	290	9
subtype1	65	81	0
subtype2	45	48	4
subtype3	21	37	1
subtype4	27	98	4
subtype5	30	26	0

Figure S80. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'MIRSEQ CHIERARCHICAL' versus 'PATHOLOGY_N_STAGE'

P value = 0.00048 (Fisher's exact test), Q value = 0.002

Table S89. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	343	78
subtype1	111	18
subtype2	70	9
subtype3	40	12
subtype4	81	36
subtype5	41	3

Figure S81. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'MIRSEQ CHIERARCHICAL' versus 'RADIATION_THERAPY'

P value = 0.0677 (Fisher's exact test), Q value = 0.13

Table S90. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	380	57
subtype1	119	13
subtype2	83	7
subtype3	43	10
subtype4	99	23
subtype5	36	4

Figure S82. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #5: 'RADIATION_THERAPY'

'MIRSEQ CHIERARCHICAL' versus 'HISTOLOGICAL_TYPE'

P value = 0.799 (Fisher's exact test), Q value = 0.83

Table S91. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	15	478
subtype1	4	144
subtype2	2	96
subtype3	3	56
subtype4	5	127
subtype5	1	55

Figure S83. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'MIRSEQ CHIERARCHICAL' versus 'RESIDUAL_TUMOR'

P value = 6e-05 (Fisher's exact test), Q value = 0.00037

Table S92. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	312	145	5	15
subtype1	109	31	0	6
subtype2	57	36	1	2
subtype3	41	14	0	2
subtype4	65	58	2	3
subtype5	40	6	2	2

Figure S84. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'MIRSEQ CHIERARCHICAL' versus 'NUMBER_OF_LYMPH_NODES'

P value = 5.3e-05 (Kruskal-Wallis (anova)), Q value = 0.00034

Table S93. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	403	0.4 (1.4)
subtype1	122	0.2 (0.8)
subtype2	77	0.2 (0.8)
subtype3	50	0.3 (0.7)
subtype4	110	1.0 (2.3)
subtype5	44	0.1 (0.4)

Figure S85. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'MIRSEQ CHIERARCHICAL' versus 'GLEASON_SCORE'

P value = 9.1e-12 (Kruskal-Wallis (anova)), Q value = 5e-10

Table S94. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	493	7.6 (1.0)
subtype1	148	7.4 (0.9)
subtype2	98	7.3 (1.1)
subtype3	59	7.6 (1.0)
subtype4	132	8.2 (1.0)
subtype5	56	7.2 (0.7)

Figure S86. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #9: 'GLEASON_SCORE'

'MIRSEQ CHIERARCHICAL' versus 'PSA_VALUE'

P value = 0.717 (Kruskal-Wallis (anova)), Q value = 0.79

Table S95. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	438	1.8 (15.9)
subtype1	137	0.9 (3.9)
subtype2	94	0.7 (3.5)
subtype3	53	1.4 (5.8)
subtype4	109	4.4 (31.1)
subtype5	45	0.9 (2.7)

Figure S87. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #10: 'PSA_VALUE'

'MIRSEQ CHIERARCHICAL' versus 'RACE'

P value = 0.708 (Fisher's exact test), Q value = 0.79

Table S96. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	2	7	146
subtype1	1	1	51
subtype2	0	3	30
subtype3	0	0	18
subtype4	0	0	9
subtype5	1	3	38

Figure S88. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #11: 'RACE'

Clustering Approach #9: 'MIRseq Mature CNMF subtypes'

Table S97. Description of clustering approach #9: 'MIRseq Mature CNMF subtypes'

Cluster Labels	1	2	3	4
Number of samples	62	107	60	90

'MIRseq Mature CNMF subtypes' versus 'Time to Death'

P value = 0.68 (logrank test), Q value = 0.77

Table S98. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	318	7	0.8 - 165.2 (28.2)
subtype1	62	1	1.0 - 102.9 (22.0)
subtype2	106	4	0.8 - 165.2 (28.5)
subtype3	60	0	0.9 - 88.2 (30.4)
subtype4	90	2	1.1 - 109.2 (29.5)

Figure S89. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'MIRseq Mature CNMF subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.0061 (Kruskal-Wallis (anova)), Q value = 0.019

Table S99. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	312	61.0 (7.0)
subtype1	59	59.7 (7.4)
subtype2	106	62.8 (6.1)
subtype3	60	61.6 (6.8)
subtype4	87	59.3 (7.4)

Figure S90. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'MIRseq Mature CNMF subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.00553 (Fisher's exact test), Q value = 0.018

Table S100. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	119	190	5
subtype1	32	29	0
subtype2	26	77	3
subtype3	22	35	1
subtype4	39	49	1

Figure S91. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'MIRseq Mature CNMF subtypes' versus 'PATHOLOGY_N_STAGE'

P value = 0.214 (Fisher's exact test), Q value = 0.32

Table S101. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	226	49
subtype1	43	10
subtype2	74	20
subtype3	40	11
subtype4	69	8

Figure S92. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'MIRseq Mature CNMF subtypes' versus 'RADIATION_THERAPY'

P value = 0.423 (Fisher's exact test), Q value = 0.54

Table S102. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	244	42
subtype1	48	7
subtype2	77	19
subtype3	48	7
subtype4	71	9

Figure S93. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

'MIRseq Mature CNMF subtypes' versus 'HISTOLOGICAL_TYPE'

P value = 0.156 (Fisher's exact test), Q value = 0.24

Table S103. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	8	311
subtype1	2	60
subtype2	5	102
subtype3	1	59
subtype4	0	90

Figure S94. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'MIRseq Mature CNMF subtypes' versus 'RESIDUAL_TUMOR'

P value = 0.00034 (Fisher's exact test), Q value = 0.0015

Table S104. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	196	98	2	11
subtype1	48	10	0	3
subtype2	53	48	0	5
subtype3	33	22	1	1
subtype4	62	18	1	2

Figure S95. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'MIRseq Mature CNMF subtypes' versus 'NUMBER_OF_LYMPH_NODES'

P value = 0.257 (Kruskal-Wallis (anova)), Q value = 0.36

Table S105. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	260	0.5 (1.6)
subtype1	50	0.2 (0.4)
subtype2	89	0.5 (1.3)
subtype3	48	0.8 (2.4)
subtype4	73	0.4 (1.7)

Figure S96. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'MIRseq Mature CNMF subtypes' versus 'GLEASON_SCORE'

P value = 4.2e-07 (Kruskal-Wallis (anova)), Q value = 9.2e-06

Table S106. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	319	7.7 (1.0)
subtype1	62	7.3 (1.0)
subtype2	107	8.0 (1.0)
subtype3	60	7.9 (1.2)
subtype4	90	7.3 (0.8)

Figure S97. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

'MIRseq Mature CNMF subtypes' versus 'PSA_VALUE'

P value = 0.0559 (Kruskal-Wallis (anova)), Q value = 0.11

Table S107. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	283	0.9 (3.3)
subtype1	59	1.2 (3.5)
subtype2	94	0.4 (1.4)
subtype3	53	1.9 (6.1)
subtype4	77	0.5 (1.5)

Figure S98. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #10: 'PSA_VALUE'

'MIRseq Mature CNMF subtypes' versus 'RACE'

P value = 0.802 (Fisher's exact test), Q value = 0.83

Table S108. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	3	91
subtype1	0	0	27
subtype2	0	0	3
subtype3	0	0	7
subtype4	1	3	54

Figure S99. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #11: 'RACE'

Clustering Approach #10: 'MIRseq Mature cHierClus subtypes'

Table S109. Description of clustering approach #10: 'MIRseq Mature cHierClus subtypes'

Cluster Labels	1	2	3	4	5	6
Number of samples	70	41	39	84	25	60

'MIRseq Mature cHierClus subtypes' versus 'Time to Death'

P value = 0.985 (logrank test), Q value = 0.99

Table S110. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	318	7	0.8 - 165.2 (28.2)
subtype1	70	1	0.9 - 102.9 (25.2)
subtype2	41	1	4.3 - 93.7 (28.7)
subtype3	39	1	1.6 - 119.4 (25.5)
subtype4	84	3	0.8 - 165.2 (29.1)
subtype5	25	0	1.0 - 68.3 (38.8)
subtype6	59	1	1.1 - 109.2 (28.5)

Figure S100. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'MIRseq Mature cHierClus subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.0386 (Kruskal-Wallis (anova)), Q value = 0.083

Table S111. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

	nPatients	Mean (Std.Dev)
ALL	312	61.0 (7.0)
subtype1	67	60.0 (6.9)
subtype2	40	59.4 (7.0)
subtype3	39	61.8 (6.5)
subtype4	84	63.1 (6.6)
subtype5	25	60.3 (7.6)
subtype6	57	60.0 (7.2)

Figure S101. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'MIRseq Mature cHierClus subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 2e-05 (Fisher's exact test), Q value = 0.00014

Table S112. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

nPatients	T2	T3	T4
ALL	119	190	5
subtype1	35	32	1
subtype2	23	17	0
subtype3	6	33	0
subtype4	19	61	3
subtype5	8	15	1
subtype6	28	32	0

Figure S102. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #3: 'PATHOLOGY_T_STAGE'

'MIRseq Mature cHierClus subtypes' versus 'PATHOLOGY_N_STAGE'

P value = 2e-05 (Fisher's exact test), Q value = 0.00014

Table S113. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

nPatients	0	1
ALL	226	49
subtype1	46	8
subtype2	31	1
subtype3	20	17
subtype4	62	15
subtype5	19	5
subtype6	48	3

Figure S103. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY_N_STAGE'

'MIRseq Mature cHierClus subtypes' versus 'RADIATION_THERAPY'

P value = 0.339 (Fisher's exact test), Q value = 0.45

Table S114. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

nPatients	NO	YES
ALL	244	42
subtype1	55	10
subtype2	37	3
subtype3	27	9
subtype4	61	11
subtype5	19	4
subtype6	45	5

Figure S104. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #5: 'RADIATION_THERAPY'

'MIRseq Mature cHierClus subtypes' versus 'HISTOLOGICAL_TYPE'

P value = 0.664 (Fisher's exact test), Q value = 0.77

Table S115. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

nPatients	PROSTATE ADENOCARCINOMA OTHER SUBTYPE	PROSTATE ADENOCARCINOMA ACINAR TYPE
ALL	8	311
subtype1	3	67
subtype2	1	40
subtype3	1	38
subtype4	3	81
subtype5	0	25
subtype6	0	60

Figure S105. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #6: 'HISTOLOGICAL_TYPE'

'MIRseq Mature cHierClus subtypes' versus 'RESIDUAL_TUMOR'

P value = 0.00012 (Fisher's exact test), Q value = 0.00066

Table S116. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

nPatients	R0	R1	R2	RX
ALL	196	98	2	11
subtype1	53	10	0	4
subtype2	27	12	1	0
subtype3	17	20	0	2
subtype4	42	39	0	2
subtype5	17	7	0	1
subtype6	40	10	1	2

Figure S106. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #7: 'RESIDUAL_TUMOR'

'MIRseq Mature cHierClus subtypes' versus 'NUMBER_OF_LYMPH_NODES'

P value = 1.54e-06 (Kruskal-Wallis (anova)), Q value = 2.4e-05

Table S117. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

	nPatients	Mean (Std.Dev)
ALL	260	0.5 (1.6)
subtype1	46	0.2 (0.7)
subtype2	31	0.0 (0.2)
subtype3	34	1.9 (3.4)
subtype4	74	0.5 (1.4)
subtype5	24	0.2 (0.4)
subtype6	51	0.1 (0.3)

Figure S107. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #8: 'NUMBER_OF_LYMPH_NODES'

'MIRseq Mature cHierClus subtypes' versus 'GLEASON_SCORE'

P value = 1.98e-11 (Kruskal-Wallis (anova)), Q value = 7.3e-10

Table S118. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

	nPatients	Mean (Std.Dev)
ALL	319	7.7 (1.0)
subtype1	70	7.5 (1.0)
subtype2	41	7.2 (1.1)
subtype3	39	8.5 (0.8)
subtype4	84	8.1 (1.0)
subtype5	25	7.2 (0.8)
subtype6	60	7.3 (0.8)

Figure S108. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #9: 'GLEASON_SCORE'

'MIRseq Mature cHierClus subtypes' versus 'PSA_VALUE'

P value = 0.56 (Kruskal-Wallis (anova)), Q value = 0.68

Table S119. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #10: 'PSA_VALUE'

	nPatients	Mean (Std.Dev)
ALL	283	0.9 (3.3)
subtype1	63	0.2 (0.5)
subtype2	40	0.4 (1.5)
subtype3	35	2.2 (5.9)
subtype4	72	1.3 (4.2)
subtype5	25	1.2 (3.8)
subtype6	48	0.3 (0.8)

Figure S109. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #10: 'PSA_VALUE'

'MIRseq Mature cHierClus subtypes' versus 'RACE'

P value = 0.804 (Fisher's exact test), Q value = 0.83

Table S120. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #11: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	3	91
subtype1	0	0	5
subtype2	0	0	11
subtype3	0	0	3
subtype4	0	0	8
subtype5	0	0	25
subtype6	1	3	39

Figure S110. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #11: 'RACE'

Methods & Data

Input

Cluster data file = /xchip/cga/gdac-prod/tcga-gdac/jobResults/GDAC_mergedClustering/PRAD-TP/20168456/PRAD-TP.mergedcluster.txt
Clinical data file = /xchip/cga/gdac-prod/tcga-gdac/jobResults/Append_Data/PRAD-TP/19775467/PRAD-TP.merged_data.txt
Number of patients = 497
Number of clustering approaches = 10
Number of selected clinical features = 11
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

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

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.

Download Results

In addition to the links below, the full results of the analysis summarized in this report can also be downloaded programmatically using firehose_get, or interactively from either the Broad GDAC website or TCGA Data Coordination Center Portal.

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] 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)

[5] 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