Correlation between aggregated molecular cancer subtypes and selected clinical features
Uveal Melanoma (Primary solid tumor)
13 July 2018  |  None
Maintainer Information
Maintained by Broad Institute GDAC (Broad Institute of MIT & Harvard)
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 7 clinical features across 80 patients, 13 significant findings detected with P value < 0.05 and Q value < 0.25.

  • CNMF clustering analysis on array-based mRNA expression data identified 5 subtypes that correlate to 'DAYS_TO_DEATH_OR_LAST_FUP' and 'PATHOLOGY_T_STAGE'.

  • Consensus hierarchical clustering analysis on array-based mRNA expression data identified 4 subtypes that correlate to 'DAYS_TO_DEATH_OR_LAST_FUP' and 'RADIATION_THERAPY'.

  • 5 subtypes identified in current cancer cohort by 'LINCRNA CNMF'. These subtypes correlate to 'DAYS_TO_DEATH_OR_LAST_FUP'.

  • 4 subtypes identified in current cancer cohort by 'LINCRNA CHIERARCHICAL'. These subtypes correlate to 'DAYS_TO_DEATH_OR_LAST_FUP'.

  • CNMF clustering analysis on array-based miR expression data identified 5 subtypes that correlate to 'DAYS_TO_DEATH_OR_LAST_FUP'.

  • Consensus hierarchical clustering analysis on array-based miR expression data identified 5 subtypes that correlate to 'DAYS_TO_DEATH_OR_LAST_FUP' and 'YEARS_TO_BIRTH'.

  • 3 subtypes identified in current cancer cohort by 'Copy Number Ratio CNMF subtypes'. These subtypes correlate to 'DAYS_TO_DEATH_OR_LAST_FUP'.

  • 6 subtypes identified in current cancer cohort by 'Copy Number Threshold CNMF subtypes'. These subtypes correlate to 'DAYS_TO_DEATH_OR_LAST_FUP'.

  • CNMF clustering analysis on methylation data identified 4 subtypes that correlate to 'DAYS_TO_DEATH_OR_LAST_FUP'.

  • 3 subtypes identified in current cancer cohort by 'METHYLATION CHIERARCHICAL'. These subtypes correlate to 'DAYS_TO_DEATH_OR_LAST_FUP'.

Results
Overview of the results

Table 1.  Get Full Table Overview of the association between subtypes identified by 10 different clustering approaches and 7 clinical features. Shown in the table are P values (Q values). Thresholded by P value < 0.05 and Q value < 0.25, 13 significant findings detected.

Clinical
Features
DAYS
TO
DEATH
OR
LAST
FUP
YEARS
TO
BIRTH
PATHOLOGIC
STAGE
PATHOLOGY
T
STAGE
PATHOLOGY
M
STAGE
GENDER RADIATION
THERAPY
Statistical Tests logrank test Kruskal-Wallis (anova) Fisher's exact test Fisher's exact test Fisher's exact test Fisher's exact test Fisher's exact test
mRNA CNMF subtypes 0.000583
(0.00453)
0.394
(0.541)
0.0529
(0.248)
0.027
(0.172)
0.287
(0.472)
0.783
(0.879)
0.386
(0.541)
mRNA cHierClus subtypes 8.91e-07
(3.12e-05)
0.291
(0.472)
0.15
(0.438)
0.309
(0.48)
0.13
(0.403)
0.791
(0.879)
0.0497
(0.248)
LINCRNA CNMF 0.0428
(0.248)
0.06
(0.262)
0.158
(0.443)
0.271
(0.472)
0.929
(0.971)
0.966
(0.994)
0.27
(0.472)
LINCRNA CHIERARCHICAL 4.57e-05
(0.000496)
0.223
(0.472)
0.287
(0.472)
0.244
(0.472)
0.25
(0.472)
0.78
(0.879)
0.183
(0.462)
miR CNMF subtypes 1.79e-06
(4.17e-05)
0.285
(0.472)
0.231
(0.472)
0.292
(0.472)
0.053
(0.248)
0.444
(0.556)
0.889
(0.943)
miR cHierClus subtypes 5.2e-06
(7.28e-05)
0.00464
(0.0325)
0.332
(0.484)
0.292
(0.472)
0.461
(0.556)
0.327
(0.484)
0.439
(0.556)
Copy Number Ratio CNMF subtypes 2.39e-06
(4.19e-05)
0.0867
(0.32)
0.201
(0.472)
0.43
(0.556)
0.0658
(0.271)
0.83
(0.908)
0.429
(0.556)
Copy Number Threshold CNMF subtypes 0.000237
(0.00207)
0.0745
(0.29)
0.682
(0.796)
0.394
(0.541)
0.112
(0.374)
0.327
(0.484)
0.297
(0.472)
Methylation CNMF subtypes 6.68e-08
(4.67e-06)
0.0928
(0.325)
0.448
(0.556)
0.642
(0.762)
0.185
(0.462)
0.246
(0.472)
0.879
(0.943)
METHYLATION CHIERARCHICAL 4.96e-05
(0.000496)
0.133
(0.403)
0.293
(0.472)
0.456
(0.556)
0.173
(0.462)
1
(1.00)
1
(1.00)
Clustering Approach #1: 'mRNA CNMF subtypes'

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

Cluster Labels 1 2 3 4 5
Number of samples 11 11 29 21 8
'mRNA CNMF subtypes' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 0.000583 (logrank test), Q value = 0.0045

Table S2.  Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

nPatients nDeath Duration Range (Median), Month
ALL 79 22 0.1 - 85.5 (26.1)
subtype1 10 5 1.3 - 49.7 (18.2)
subtype2 11 4 0.4 - 61.2 (31.1)
subtype3 29 1 0.2 - 85.5 (27.6)
subtype4 21 7 1.4 - 82.2 (21.0)
subtype5 8 5 0.1 - 45.3 (23.8)

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

'mRNA CNMF subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.394 (Kruskal-Wallis (anova)), Q value = 0.54

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

nPatients Mean (Std.Dev)
ALL 79 61.5 (14.0)
subtype1 10 67.4 (15.6)
subtype2 11 59.8 (14.1)
subtype3 29 57.8 (15.0)
subtype4 21 64.7 (12.8)
subtype5 8 62.1 (8.7)

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

'mRNA CNMF subtypes' versus 'PATHOLOGIC_STAGE'

P value = 0.0529 (Fisher's exact test), Q value = 0.25

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

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 11 27 25 10 1 4
subtype1 1 2 3 2 1 1
subtype2 0 3 4 3 0 0
subtype3 9 12 6 2 0 0
subtype4 0 7 9 3 0 2
subtype5 1 3 3 0 0 1

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

'mRNA CNMF subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.027 (Kruskal-Wallis (anova)), Q value = 0.17

Table S5.  Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 13 32 34
subtype1 1 3 6
subtype2 2 5 4
subtype3 9 14 6
subtype4 0 7 14
subtype5 1 3 4

Figure S4.  Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

'mRNA CNMF subtypes' versus 'PATHOLOGY_M_STAGE'

P value = 0.287 (Fisher's exact test), Q value = 0.47

Table S6.  Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 51 4
subtype1 8 1
subtype2 8 0
subtype3 17 0
subtype4 14 2
subtype5 4 1

Figure S5.  Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'mRNA CNMF subtypes' versus 'GENDER'

P value = 0.783 (Fisher's exact test), Q value = 0.88

Table S7.  Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #6: 'GENDER'

nPatients FEMALE MALE
ALL 34 45
subtype1 6 4
subtype2 4 7
subtype3 12 17
subtype4 8 13
subtype5 4 4

Figure S6.  Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #6: 'GENDER'

'mRNA CNMF subtypes' versus 'RADIATION_THERAPY'

P value = 0.386 (Fisher's exact test), Q value = 0.54

Table S8.  Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 75 3
subtype1 8 1
subtype2 10 1
subtype3 28 1
subtype4 21 0
subtype5 8 0

Figure S7.  Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

Clustering Approach #2: 'mRNA cHierClus subtypes'

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

Cluster Labels 1 2 3 4
Number of samples 15 15 27 23
'mRNA cHierClus subtypes' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 8.91e-07 (logrank test), Q value = 3.1e-05

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

nPatients nDeath Duration Range (Median), Month
ALL 79 22 0.1 - 85.5 (26.1)
subtype1 15 9 0.1 - 45.3 (18.9)
subtype2 15 2 0.4 - 61.2 (27.6)
subtype3 27 1 0.2 - 85.5 (38.7)
subtype4 22 10 1.3 - 49.7 (19.8)

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

'mRNA cHierClus subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.291 (Kruskal-Wallis (anova)), Q value = 0.47

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

nPatients Mean (Std.Dev)
ALL 79 61.5 (14.0)
subtype1 15 63.2 (12.9)
subtype2 15 60.9 (14.0)
subtype3 27 57.4 (14.7)
subtype4 22 65.9 (13.3)

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

'mRNA cHierClus subtypes' versus 'PATHOLOGIC_STAGE'

P value = 0.15 (Fisher's exact test), Q value = 0.44

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

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 11 27 25 10 1 4
subtype1 1 6 4 1 1 2
subtype2 1 5 4 4 0 0
subtype3 8 10 7 2 0 0
subtype4 1 6 10 3 0 2

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

'mRNA cHierClus subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.309 (Kruskal-Wallis (anova)), Q value = 0.48

Table S13.  Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 13 32 34
subtype1 2 5 8
subtype2 2 7 6
subtype3 8 11 8
subtype4 1 9 12

Figure S11.  Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

'mRNA cHierClus subtypes' versus 'PATHOLOGY_M_STAGE'

P value = 0.13 (Fisher's exact test), Q value = 0.4

Table S14.  Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 51 4
subtype1 9 2
subtype2 10 0
subtype3 18 0
subtype4 14 2

Figure S12.  Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'mRNA cHierClus subtypes' versus 'GENDER'

P value = 0.791 (Fisher's exact test), Q value = 0.88

Table S15.  Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #6: 'GENDER'

nPatients FEMALE MALE
ALL 34 45
subtype1 8 7
subtype2 6 9
subtype3 12 15
subtype4 8 14

Figure S13.  Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #6: 'GENDER'

'mRNA cHierClus subtypes' versus 'RADIATION_THERAPY'

P value = 0.0497 (Fisher's exact test), Q value = 0.25

Table S16.  Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 75 3
subtype1 13 1
subtype2 13 2
subtype3 27 0
subtype4 22 0

Figure S14.  Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

Clustering Approach #3: 'LINCRNA CNMF'

Table S17.  Description of clustering approach #3: 'LINCRNA CNMF'

Cluster Labels 1 2 3 4 5
Number of samples 18 7 28 11 16
'LINCRNA CNMF' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 0.0428 (logrank test), Q value = 0.25

Table S18.  Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

nPatients nDeath Duration Range (Median), Month
ALL 79 22 0.1 - 85.5 (26.1)
subtype1 18 10 0.1 - 61.2 (24.7)
subtype2 7 1 0.2 - 27.0 (15.1)
subtype3 28 7 1.4 - 82.2 (24.3)
subtype4 10 3 1.3 - 85.5 (39.7)
subtype5 16 1 0.6 - 52.0 (34.4)

Figure S15.  Get High-res Image Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

'LINCRNA CNMF' versus 'YEARS_TO_BIRTH'

P value = 0.06 (Kruskal-Wallis (anova)), Q value = 0.26

Table S19.  Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

nPatients Mean (Std.Dev)
ALL 79 61.5 (14.0)
subtype1 18 62.9 (11.4)
subtype2 7 61.3 (15.0)
subtype3 28 64.2 (12.9)
subtype4 10 67.0 (14.3)
subtype5 16 52.0 (15.0)

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

'LINCRNA CNMF' versus 'PATHOLOGIC_STAGE'

P value = 0.158 (Fisher's exact test), Q value = 0.44

Table S20.  Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 11 27 25 10 1 4
subtype1 1 5 7 2 1 1
subtype2 2 3 0 2 0 0
subtype3 1 9 12 4 0 2
subtype4 2 3 2 2 0 1
subtype5 5 7 4 0 0 0

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

'LINCRNA CNMF' versus 'PATHOLOGY_T_STAGE'

P value = 0.271 (Kruskal-Wallis (anova)), Q value = 0.47

Table S21.  Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 13 32 34
subtype1 2 7 9
subtype2 2 3 2
subtype3 2 10 16
subtype4 2 4 4
subtype5 5 8 3

Figure S18.  Get High-res Image Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

'LINCRNA CNMF' versus 'PATHOLOGY_M_STAGE'

P value = 0.929 (Fisher's exact test), Q value = 0.97

Table S22.  Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 51 4
subtype1 12 1
subtype2 5 0
subtype3 20 2
subtype4 6 1
subtype5 8 0

Figure S19.  Get High-res Image Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'LINCRNA CNMF' versus 'GENDER'

P value = 0.966 (Fisher's exact test), Q value = 0.99

Table S23.  Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #6: 'GENDER'

nPatients FEMALE MALE
ALL 34 45
subtype1 9 9
subtype2 3 4
subtype3 12 16
subtype4 4 6
subtype5 6 10

Figure S20.  Get High-res Image Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #6: 'GENDER'

'LINCRNA CNMF' versus 'RADIATION_THERAPY'

P value = 0.27 (Fisher's exact test), Q value = 0.47

Table S24.  Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 75 3
subtype1 15 2
subtype2 7 0
subtype3 28 0
subtype4 10 0
subtype5 15 1

Figure S21.  Get High-res Image Clustering Approach #3: 'LINCRNA CNMF' versus Clinical Feature #7: 'RADIATION_THERAPY'

Clustering Approach #4: 'LINCRNA CHIERARCHICAL'

Table S25.  Description of clustering approach #4: 'LINCRNA CHIERARCHICAL'

Cluster Labels 1 2 3 4
Number of samples 21 24 20 15
'LINCRNA CHIERARCHICAL' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 4.57e-05 (logrank test), Q value = 5e-04

Table S26.  Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

nPatients nDeath Duration Range (Median), Month
ALL 79 22 0.1 - 85.5 (26.1)
subtype1 20 11 0.1 - 61.2 (23.7)
subtype2 24 1 0.2 - 82.2 (24.3)
subtype3 20 9 1.4 - 45.9 (19.8)
subtype4 15 1 20.8 - 85.5 (39.8)

Figure S22.  Get High-res Image Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

'LINCRNA CHIERARCHICAL' versus 'YEARS_TO_BIRTH'

P value = 0.223 (Kruskal-Wallis (anova)), Q value = 0.47

Table S27.  Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

nPatients Mean (Std.Dev)
ALL 79 61.5 (14.0)
subtype1 20 64.8 (12.5)
subtype2 24 59.6 (13.4)
subtype3 20 65.2 (13.2)
subtype4 15 55.4 (16.4)

Figure S23.  Get High-res Image Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'LINCRNA CHIERARCHICAL' versus 'PATHOLOGIC_STAGE'

P value = 0.287 (Fisher's exact test), Q value = 0.47

Table S28.  Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 11 27 25 10 1 4
subtype1 1 6 6 3 1 2
subtype2 5 8 6 5 0 0
subtype3 1 6 9 2 0 2
subtype4 4 7 4 0 0 0

Figure S24.  Get High-res Image Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

'LINCRNA CHIERARCHICAL' versus 'PATHOLOGY_T_STAGE'

P value = 0.244 (Kruskal-Wallis (anova)), Q value = 0.47

Table S29.  Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 13 32 34
subtype1 2 8 10
subtype2 6 8 10
subtype3 1 8 11
subtype4 4 8 3

Figure S25.  Get High-res Image Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

'LINCRNA CHIERARCHICAL' versus 'PATHOLOGY_M_STAGE'

P value = 0.25 (Fisher's exact test), Q value = 0.47

Table S30.  Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 51 4
subtype1 12 2
subtype2 19 0
subtype3 12 2
subtype4 8 0

Figure S26.  Get High-res Image Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'LINCRNA CHIERARCHICAL' versus 'GENDER'

P value = 0.78 (Fisher's exact test), Q value = 0.88

Table S31.  Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #6: 'GENDER'

nPatients FEMALE MALE
ALL 34 45
subtype1 9 11
subtype2 12 12
subtype3 7 13
subtype4 6 9

Figure S27.  Get High-res Image Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #6: 'GENDER'

'LINCRNA CHIERARCHICAL' versus 'RADIATION_THERAPY'

P value = 0.183 (Fisher's exact test), Q value = 0.46

Table S32.  Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 75 3
subtype1 17 2
subtype2 24 0
subtype3 20 0
subtype4 14 1

Figure S28.  Get High-res Image Clustering Approach #4: 'LINCRNA CHIERARCHICAL' versus Clinical Feature #7: 'RADIATION_THERAPY'

Clustering Approach #5: 'miR CNMF subtypes'

Table S33.  Description of clustering approach #5: 'miR CNMF subtypes'

Cluster Labels 1 2 3 4 5
Number of samples 13 21 18 24 4
'miR CNMF subtypes' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 1.79e-06 (logrank test), Q value = 4.2e-05

Table S34.  Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

nPatients nDeath Duration Range (Median), Month
ALL 79 22 0.1 - 85.5 (26.1)
subtype1 13 6 0.4 - 61.2 (28.7)
subtype2 21 12 0.1 - 45.3 (18.9)
subtype3 17 2 0.2 - 85.5 (26.2)
subtype4 24 0 0.2 - 82.2 (32.8)
subtype5 4 2 1.3 - 27.0 (14.1)

Figure S29.  Get High-res Image Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

'miR CNMF subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.285 (Kruskal-Wallis (anova)), Q value = 0.47

Table S35.  Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

nPatients Mean (Std.Dev)
ALL 79 61.5 (14.0)
subtype1 13 60.4 (15.9)
subtype2 21 65.2 (11.6)
subtype3 17 61.2 (12.6)
subtype4 24 57.4 (15.9)
subtype5 4 72.2 (3.1)

Figure S30.  Get High-res Image Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'miR CNMF subtypes' versus 'PATHOLOGIC_STAGE'

P value = 0.231 (Fisher's exact test), Q value = 0.47

Table S36.  Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 11 27 25 10 1 4
subtype1 0 3 5 2 1 1
subtype2 1 8 7 3 0 2
subtype3 2 8 6 1 0 0
subtype4 7 7 7 3 0 0
subtype5 1 1 0 1 0 1

Figure S31.  Get High-res Image Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

'miR CNMF subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.292 (Kruskal-Wallis (anova)), Q value = 0.47

Table S37.  Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 13 32 34
subtype1 0 7 6
subtype2 2 7 12
subtype3 3 9 5
subtype4 7 8 9
subtype5 1 1 2

Figure S32.  Get High-res Image Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

'miR CNMF subtypes' versus 'PATHOLOGY_M_STAGE'

P value = 0.053 (Fisher's exact test), Q value = 0.25

Table S38.  Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 51 4
subtype1 9 1
subtype2 12 2
subtype3 13 0
subtype4 16 0
subtype5 1 1

Figure S33.  Get High-res Image Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'miR CNMF subtypes' versus 'GENDER'

P value = 0.444 (Fisher's exact test), Q value = 0.56

Table S39.  Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #6: 'GENDER'

nPatients FEMALE MALE
ALL 34 45
subtype1 6 7
subtype2 6 15
subtype3 10 7
subtype4 10 14
subtype5 2 2

Figure S34.  Get High-res Image Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #6: 'GENDER'

'miR CNMF subtypes' versus 'RADIATION_THERAPY'

P value = 0.889 (Fisher's exact test), Q value = 0.94

Table S40.  Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 75 3
subtype1 11 1
subtype2 20 1
subtype3 17 0
subtype4 23 1
subtype5 4 0

Figure S35.  Get High-res Image Clustering Approach #5: 'miR CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

Clustering Approach #6: 'miR cHierClus subtypes'

Table S41.  Description of clustering approach #6: 'miR cHierClus subtypes'

Cluster Labels 1 2 3 4 5
Number of samples 21 20 8 17 14
'miR cHierClus subtypes' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 5.2e-06 (logrank test), Q value = 7.3e-05

Table S42.  Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

nPatients nDeath Duration Range (Median), Month
ALL 79 22 0.1 - 85.5 (26.1)
subtype1 20 12 0.1 - 49.7 (13.4)
subtype2 20 7 2.1 - 61.2 (32.0)
subtype3 8 1 0.2 - 85.5 (23.2)
subtype4 17 1 0.2 - 82.2 (35.0)
subtype5 14 1 0.4 - 44.3 (27.0)

Figure S36.  Get High-res Image Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

'miR cHierClus subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.00464 (Kruskal-Wallis (anova)), Q value = 0.032

Table S43.  Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

nPatients Mean (Std.Dev)
ALL 79 61.5 (14.0)
subtype1 20 68.6 (13.4)
subtype2 20 61.0 (9.9)
subtype3 8 59.8 (14.5)
subtype4 17 64.1 (15.8)
subtype5 14 50.2 (11.1)

Figure S37.  Get High-res Image Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'miR cHierClus subtypes' versus 'PATHOLOGIC_STAGE'

P value = 0.332 (Fisher's exact test), Q value = 0.48

Table S44.  Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 11 27 25 10 1 4
subtype1 1 8 6 1 1 3
subtype2 1 6 7 4 0 1
subtype3 0 5 2 1 0 0
subtype4 5 5 4 3 0 0
subtype5 4 3 6 1 0 0

Figure S38.  Get High-res Image Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

'miR cHierClus subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.292 (Kruskal-Wallis (anova)), Q value = 0.47

Table S45.  Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 13 32 34
subtype1 1 10 9
subtype2 3 7 10
subtype3 0 5 3
subtype4 5 7 5
subtype5 4 3 7

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

'miR cHierClus subtypes' versus 'PATHOLOGY_M_STAGE'

P value = 0.461 (Fisher's exact test), Q value = 0.56

Table S46.  Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 51 4
subtype1 14 3
subtype2 10 1
subtype3 6 0
subtype4 11 0
subtype5 10 0

Figure S40.  Get High-res Image Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'miR cHierClus subtypes' versus 'GENDER'

P value = 0.327 (Fisher's exact test), Q value = 0.48

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

nPatients FEMALE MALE
ALL 34 45
subtype1 11 9
subtype2 5 15
subtype3 3 5
subtype4 9 8
subtype5 6 8

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

'miR cHierClus subtypes' versus 'RADIATION_THERAPY'

P value = 0.439 (Fisher's exact test), Q value = 0.56

Table S48.  Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 75 3
subtype1 19 0
subtype2 18 2
subtype3 8 0
subtype4 17 0
subtype5 13 1

Figure S42.  Get High-res Image Clustering Approach #6: 'miR cHierClus subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

Clustering Approach #7: 'Copy Number Ratio CNMF subtypes'

Table S49.  Description of clustering approach #7: 'Copy Number Ratio CNMF subtypes'

Cluster Labels 1 2 3 4 5
Number of samples 18 21 37 2 2
'Copy Number Ratio CNMF subtypes' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 2.39e-06 (logrank test), Q value = 4.2e-05

Table S50.  Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

nPatients nDeath Duration Range (Median), Month
ALL 75 22 0.1 - 85.5 (25.0)
subtype1 17 9 1.6 - 49.7 (19.6)
subtype2 21 11 0.1 - 45.3 (21.0)
subtype3 36 2 0.2 - 85.5 (29.7)

Figure S43.  Get High-res Image Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

'Copy Number Ratio CNMF subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.0867 (Kruskal-Wallis (anova)), Q value = 0.32

Table S51.  Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

nPatients Mean (Std.Dev)
ALL 75 61.6 (14.1)
subtype1 17 64.3 (14.0)
subtype2 21 66.3 (12.0)
subtype3 37 57.6 (14.5)

Figure S44.  Get High-res Image Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'Copy Number Ratio CNMF subtypes' versus 'PATHOLOGIC_STAGE'

P value = 0.201 (Fisher's exact test), Q value = 0.47

Table S52.  Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 11 27 22 9 1 4
subtype1 1 5 7 2 0 2
subtype2 1 8 5 3 1 2
subtype3 9 14 10 4 0 0

Figure S45.  Get High-res Image Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

'Copy Number Ratio CNMF subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.43 (Kruskal-Wallis (anova)), Q value = 0.56

Table S53.  Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 13 30 32
subtype1 2 6 9
subtype2 2 8 11
subtype3 9 16 12

Figure S46.  Get High-res Image Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

'Copy Number Ratio CNMF subtypes' versus 'PATHOLOGY_M_STAGE'

P value = 0.0658 (Fisher's exact test), Q value = 0.27

Table S54.  Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 50 4
subtype1 12 2
subtype2 12 2
subtype3 26 0

Figure S47.  Get High-res Image Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'Copy Number Ratio CNMF subtypes' versus 'GENDER'

P value = 0.83 (Fisher's exact test), Q value = 0.91

Table S55.  Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #6: 'GENDER'

nPatients FEMALE MALE
ALL 34 41
subtype1 9 8
subtype2 9 12
subtype3 16 21

Figure S48.  Get High-res Image Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #6: 'GENDER'

'Copy Number Ratio CNMF subtypes' versus 'RADIATION_THERAPY'

P value = 0.429 (Fisher's exact test), Q value = 0.56

Table S56.  Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 71 3
subtype1 16 0
subtype2 19 2
subtype3 36 1

Figure S49.  Get High-res Image Clustering Approach #7: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

Clustering Approach #8: 'Copy Number Threshold CNMF subtypes'

Table S57.  Description of clustering approach #8: 'Copy Number Threshold CNMF subtypes'

Cluster Labels 1 2 3 4 5 6 7 8
Number of samples 16 7 22 12 2 14 5 2
'Copy Number Threshold CNMF subtypes' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 0.000237 (logrank test), Q value = 0.0021

Table S58.  Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

nPatients nDeath Duration Range (Median), Month
ALL 75 21 0.1 - 85.5 (25.0)
subtype1 15 7 1.6 - 49.7 (19.6)
subtype2 7 3 1.4 - 32.3 (13.6)
subtype3 21 2 0.2 - 85.5 (26.1)
subtype4 12 0 0.6 - 82.2 (36.4)
subtype6 14 6 0.1 - 61.2 (36.9)
subtype7 5 3 2.2 - 36.6 (24.0)

Figure S50.  Get High-res Image Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

'Copy Number Threshold CNMF subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.0745 (Kruskal-Wallis (anova)), Q value = 0.29

Table S59.  Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

nPatients Mean (Std.Dev)
ALL 75 61.7 (14.0)
subtype1 15 64.3 (16.3)
subtype2 7 73.4 (11.3)
subtype3 22 61.2 (14.4)
subtype4 12 55.2 (11.3)
subtype6 14 59.1 (11.8)
subtype7 5 63.4 (13.9)

Figure S51.  Get High-res Image Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'Copy Number Threshold CNMF subtypes' versus 'PATHOLOGIC_STAGE'

P value = 0.682 (Fisher's exact test), Q value = 0.8

Table S60.  Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 10 26 23 10 1 4
subtype1 1 5 5 2 0 2
subtype2 0 2 2 1 1 1
subtype3 5 8 6 3 0 0
subtype4 3 4 3 2 0 0
subtype6 0 6 5 2 0 0
subtype7 1 1 2 0 0 1

Figure S52.  Get High-res Image Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

'Copy Number Threshold CNMF subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.394 (Kruskal-Wallis (anova)), Q value = 0.54

Table S61.  Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 12 31 32
subtype1 2 6 7
subtype2 1 1 5
subtype3 5 10 7
subtype4 3 4 5
subtype6 0 9 5
subtype7 1 1 3

Figure S53.  Get High-res Image Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

'Copy Number Threshold CNMF subtypes' versus 'PATHOLOGY_M_STAGE'

P value = 0.112 (Fisher's exact test), Q value = 0.37

Table S62.  Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 48 4
subtype1 11 2
subtype2 5 1
subtype3 16 0
subtype4 8 0
subtype6 6 0
subtype7 2 1

Figure S54.  Get High-res Image Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'Copy Number Threshold CNMF subtypes' versus 'GENDER'

P value = 0.327 (Fisher's exact test), Q value = 0.48

Table S63.  Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #6: 'GENDER'

nPatients FEMALE MALE
ALL 32 43
subtype1 6 9
subtype2 1 6
subtype3 11 11
subtype4 4 8
subtype6 6 8
subtype7 4 1

Figure S55.  Get High-res Image Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #6: 'GENDER'

'Copy Number Threshold CNMF subtypes' versus 'RADIATION_THERAPY'

P value = 0.297 (Fisher's exact test), Q value = 0.47

Table S64.  Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 72 2
subtype1 14 0
subtype2 6 1
subtype3 22 0
subtype4 12 0
subtype6 13 1
subtype7 5 0

Figure S56.  Get High-res Image Clustering Approach #8: 'Copy Number Threshold CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

Clustering Approach #9: 'Methylation CNMF subtypes'

Table S65.  Description of clustering approach #9: 'Methylation CNMF subtypes'

Cluster Labels 1 2 3 4
Number of samples 28 13 16 23
'Methylation CNMF subtypes' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 6.68e-08 (logrank test), Q value = 4.7e-06

Table S66.  Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

nPatients nDeath Duration Range (Median), Month
ALL 79 22 0.1 - 85.5 (26.1)
subtype1 28 17 0.1 - 45.3 (19.3)
subtype2 13 1 0.2 - 85.5 (27.5)
subtype3 15 2 1.3 - 82.2 (31.8)
subtype4 23 2 0.2 - 46.8 (27.6)

Figure S57.  Get High-res Image Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

'Methylation CNMF subtypes' versus 'YEARS_TO_BIRTH'

P value = 0.0928 (Kruskal-Wallis (anova)), Q value = 0.32

Table S67.  Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

nPatients Mean (Std.Dev)
ALL 79 61.5 (14.0)
subtype1 28 66.9 (12.1)
subtype2 13 60.8 (12.9)
subtype3 15 60.0 (13.2)
subtype4 23 56.4 (15.8)

Figure S58.  Get High-res Image Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'Methylation CNMF subtypes' versus 'PATHOLOGIC_STAGE'

P value = 0.448 (Fisher's exact test), Q value = 0.56

Table S68.  Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 11 27 25 10 1 4
subtype1 1 10 8 3 1 4
subtype2 2 4 5 2 0 0
subtype3 2 4 7 2 0 0
subtype4 6 9 5 3 0 0

Figure S59.  Get High-res Image Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

'Methylation CNMF subtypes' versus 'PATHOLOGY_T_STAGE'

P value = 0.642 (Kruskal-Wallis (anova)), Q value = 0.76

Table S69.  Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 13 32 34
subtype1 2 11 15
subtype2 2 6 5
subtype3 3 6 6
subtype4 6 9 8

Figure S60.  Get High-res Image Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

'Methylation CNMF subtypes' versus 'PATHOLOGY_M_STAGE'

P value = 0.185 (Fisher's exact test), Q value = 0.46

Table S70.  Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 51 4
subtype1 18 4
subtype2 6 0
subtype3 11 0
subtype4 16 0

Figure S61.  Get High-res Image Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'Methylation CNMF subtypes' versus 'GENDER'

P value = 0.246 (Fisher's exact test), Q value = 0.47

Table S71.  Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #6: 'GENDER'

nPatients FEMALE MALE
ALL 34 45
subtype1 11 17
subtype2 5 8
subtype3 10 5
subtype4 8 15

Figure S62.  Get High-res Image Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #6: 'GENDER'

'Methylation CNMF subtypes' versus 'RADIATION_THERAPY'

P value = 0.879 (Fisher's exact test), Q value = 0.94

Table S72.  Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 75 3
subtype1 25 2
subtype2 13 0
subtype3 15 0
subtype4 22 1

Figure S63.  Get High-res Image Clustering Approach #9: 'Methylation CNMF subtypes' versus Clinical Feature #7: 'RADIATION_THERAPY'

Clustering Approach #10: 'METHYLATION CHIERARCHICAL'

Table S73.  Description of clustering approach #10: 'METHYLATION CHIERARCHICAL'

Cluster Labels 1 2 3
Number of samples 40 15 25
'METHYLATION CHIERARCHICAL' versus 'DAYS_TO_DEATH_OR_LAST_FUP'

P value = 4.96e-05 (logrank test), Q value = 5e-04

Table S74.  Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

nPatients nDeath Duration Range (Median), Month
ALL 79 22 0.1 - 85.5 (26.1)
subtype1 39 19 0.1 - 61.2 (19.9)
subtype2 15 2 0.2 - 85.5 (31.8)
subtype3 25 1 0.2 - 46.8 (27.5)

Figure S64.  Get High-res Image Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

'METHYLATION CHIERARCHICAL' versus 'YEARS_TO_BIRTH'

P value = 0.133 (Kruskal-Wallis (anova)), Q value = 0.4

Table S75.  Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

nPatients Mean (Std.Dev)
ALL 79 61.5 (14.0)
subtype1 39 64.9 (12.8)
subtype2 15 60.9 (13.0)
subtype3 25 56.7 (15.4)

Figure S65.  Get High-res Image Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #2: 'YEARS_TO_BIRTH'

'METHYLATION CHIERARCHICAL' versus 'PATHOLOGIC_STAGE'

P value = 0.293 (Fisher's exact test), Q value = 0.47

Table S76.  Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

nPatients STAGE IIA STAGE IIB STAGE IIIA STAGE IIIB STAGE IIIC STAGE IV
ALL 11 27 25 10 1 4
subtype1 2 13 14 4 1 4
subtype2 2 5 5 3 0 0
subtype3 7 9 6 3 0 0

Figure S66.  Get High-res Image Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #3: 'PATHOLOGIC_STAGE'

'METHYLATION CHIERARCHICAL' versus 'PATHOLOGY_T_STAGE'

P value = 0.456 (Kruskal-Wallis (anova)), Q value = 0.56

Table S77.  Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

T2 T3 T4
ALL 13 32 34
subtype1 4 16 19
subtype2 2 7 6
subtype3 7 9 9

Figure S67.  Get High-res Image Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #4: 'PATHOLOGY_T_STAGE'

'METHYLATION CHIERARCHICAL' versus 'PATHOLOGY_M_STAGE'

P value = 0.173 (Fisher's exact test), Q value = 0.46

Table S78.  Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

nPatients CLASS0 CLASS1
ALL 51 4
subtype1 24 4
subtype2 11 0
subtype3 16 0

Figure S68.  Get High-res Image Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #5: 'PATHOLOGY_M_STAGE'

'METHYLATION CHIERARCHICAL' versus 'GENDER'

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

Table S79.  Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #6: 'GENDER'

nPatients FEMALE MALE
ALL 34 45
subtype1 17 22
subtype2 6 9
subtype3 11 14

Figure S69.  Get High-res Image Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #6: 'GENDER'

'METHYLATION CHIERARCHICAL' versus 'RADIATION_THERAPY'

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

Table S80.  Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #7: 'RADIATION_THERAPY'

nPatients NO YES
ALL 75 3
subtype1 36 2
subtype2 15 0
subtype3 24 1

Figure S70.  Get High-res Image Clustering Approach #10: 'METHYLATION CHIERARCHICAL' versus Clinical Feature #7: 'RADIATION_THERAPY'

Methods & Data
Input
  • Cluster data file = /cromwell_root/fc-f5144117-2d5a-42c2-8998-5b38e52db5d9/9d897bdb-7a9a-423a-ad17-06f28ae98803/aggregate_clusters_workflow/1d8bac82-5a1d-4cd2-aeb1-8ddd0e1e7a20/call-aggregate_clusters/TCGA-UVM-TP.mergedcluster.txt

  • Clinical data file = /cromwell_root/fc-2289d790-de74-4808-9b0a-cefafc34d859/0d7c7dcf-18e0-4b2d-afc0-a0b2ee1e45ff/preprocess_clinical_workflow/70152ac6-f707-4277-8d60-8770b1b366c6/call-preprocess_clinical/TCGA-UVM-TP.clin.merged.picked.txt

  • Number of patients = 80

  • Number of clustering approaches = 10

  • Number of selected clinical features = 7

  • 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.

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