Correlation between aggregated molecular cancer subtypes and selected clinical features

Overview

Introduction

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

Summary

Testing the association between subtypes identified by 7 different clustering approaches and 4 clinical features across 33 patients, no significant finding detected with P value < 0.05 and Q value < 0.25.

2 subtypes identified in current cancer cohort by 'METHLYATION CNMF'. These subtypes do not correlate to any clinical features.
CNMF clustering analysis on sequencing-based mRNA expression data identified 2 subtypes that do not correlate to any clinical features.
Consensus hierarchical clustering analysis on sequencing-based mRNA expression data identified 3 subtypes that do not correlate to any clinical features.
2 subtypes identified in current cancer cohort by 'MIRSEQ CNMF'. These subtypes do not correlate to any clinical features.
5 subtypes identified in current cancer cohort by 'MIRSEQ CHIERARCHICAL'. These subtypes do not correlate to any clinical features.
3 subtypes identified in current cancer cohort by 'MIRseq Mature CNMF subtypes'. These subtypes do not correlate to any clinical features.
5 subtypes identified in current cancer cohort by 'MIRseq Mature cHierClus subtypes'. These subtypes do not correlate to any clinical features.

Results

Overview of the results

Table 1. Get Full Table Overview of the association between subtypes identified by 7 different clustering approaches and 4 clinical features. Shown in the table are P values (Q values). Thresholded by P value < 0.05 and Q value < 0.25, no significant finding detected.


Clinical Features	Time to Death	AGE	GENDER	RACE
Statistical Tests	logrank test	Kruskal-Wallis (anova)	Fisher's exact test	Fisher's exact test
METHLYATION CNMF	0.205 (1.00)	0.428 (1.00)	0.732 (1.00)	0.0525 (1.00)
RNAseq CNMF subtypes	0.0302 (0.845)	0.288 (1.00)	0.111 (1.00)	0.308 (1.00)
RNAseq cHierClus subtypes	0.0302 (0.845)	0.559 (1.00)	0.482 (1.00)	0.776 (1.00)
MIRSEQ CNMF	0.0458 (1.00)	0.954 (1.00)	0.682 (1.00)	0.825 (1.00)
MIRSEQ CHIERARCHICAL	0.443 (1.00)	0.939 (1.00)	0.623 (1.00)	0.146 (1.00)
MIRseq Mature CNMF subtypes	0.434 (1.00)	0.892 (1.00)	0.512 (1.00)	0.907 (1.00)
MIRseq Mature cHierClus subtypes	0.428 (1.00)	0.861 (1.00)	0.769 (1.00)	0.939 (1.00)

Clustering Approach #1: 'METHLYATION CNMF'

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

Cluster Labels	1	2
Number of samples	16	17

'METHLYATION CNMF' versus 'Time to Death'

P value = 0.205 (logrank test), Q value = 1

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

	nPatients	nDeath	Duration Range (Median), Month
ALL	32	5	0.2 - 211.2 (25.4)
subtype1	15	1	0.2 - 196.6 (26.7)
subtype2	17	4	0.4 - 211.2 (15.0)

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

'METHLYATION CNMF' versus 'AGE'

P value = 0.428 (Wilcoxon-test), Q value = 1

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

	nPatients	Mean (Std.Dev)
ALL	33	57.2 (12.2)
subtype1	16	59.3 (11.3)
subtype2	17	55.1 (13.0)

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

'METHLYATION CNMF' versus 'GENDER'

P value = 0.732 (Fisher's exact test), Q value = 1

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

nPatients	FEMALE	MALE
ALL	18	15
subtype1	8	8
subtype2	10	7

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

'METHLYATION CNMF' versus 'RACE'

P value = 0.0525 (Fisher's exact test), Q value = 1

Table S5. Clustering Approach #1: 'METHLYATION CNMF' versus Clinical Feature #4: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	17	1	15
subtype1	5	1	10
subtype2	12	0	5

Figure S4. Get High-res Image Clustering Approach #1: 'METHLYATION CNMF' versus Clinical Feature #4: 'RACE'

Clustering Approach #2: 'RNAseq CNMF subtypes'

Table S6. Description of clustering approach #2: 'RNAseq CNMF subtypes'

Cluster Labels	1	2
Number of samples	13	12

'RNAseq CNMF subtypes' versus 'Time to Death'

P value = 0.0302 (logrank test), Q value = 0.84

Table S7. Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	25	4	0.2 - 211.2 (29.9)
subtype1	13	2	4.3 - 211.2 (46.3)
subtype2	12	2	0.2 - 74.1 (20.9)

Figure S5. Get High-res Image Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq CNMF subtypes' versus 'AGE'

P value = 0.288 (Wilcoxon-test), Q value = 1

Table S8. Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	25	56.2 (12.8)
subtype1	13	59.7 (10.7)
subtype2	12	52.5 (14.3)

Figure S6. Get High-res Image Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #2: 'AGE'

'RNAseq CNMF subtypes' versus 'GENDER'

P value = 0.111 (Fisher's exact test), Q value = 1

Table S9. Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	14	11
subtype1	5	8
subtype2	9	3

Figure S7. Get High-res Image Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'GENDER'

'RNAseq CNMF subtypes' versus 'RACE'

P value = 0.308 (Fisher's exact test), Q value = 1

Table S10. Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #4: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	10	1	14
subtype1	4	0	9
subtype2	6	1	5

Figure S8. Get High-res Image Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #4: 'RACE'

Clustering Approach #3: 'RNAseq cHierClus subtypes'

Table S11. Description of clustering approach #3: 'RNAseq cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	12	9	4

'RNAseq cHierClus subtypes' versus 'Time to Death'

P value = 0.0302 (logrank test), Q value = 0.84

Table S12. Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	25	4	0.2 - 211.2 (29.9)
subtype1	12	2	4.3 - 211.2 (41.5)
subtype2	9	2	0.2 - 74.1 (4.6)
subtype3	4	0	15.0 - 46.3 (25.7)

Figure S9. Get High-res Image Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq cHierClus subtypes' versus 'AGE'

P value = 0.559 (Kruskal-Wallis (anova)), Q value = 1

Table S13. Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	25	56.2 (12.8)
subtype1	12	58.4 (13.2)
subtype2	9	55.1 (13.7)
subtype3	4	52.2 (11.6)

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

'RNAseq cHierClus subtypes' versus 'GENDER'

P value = 0.482 (Fisher's exact test), Q value = 1

Table S14. Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	14	11
subtype1	5	7
subtype2	6	3
subtype3	3	1

Figure S11. Get High-res Image Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

'RNAseq cHierClus subtypes' versus 'RACE'

P value = 0.776 (Fisher's exact test), Q value = 1

Table S15. Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #4: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	10	1	14
subtype1	5	0	7
subtype2	4	1	4
subtype3	1	0	3

Figure S12. Get High-res Image Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #4: 'RACE'

Clustering Approach #4: 'MIRSEQ CNMF'

Table S16. Description of clustering approach #4: 'MIRSEQ CNMF'

Cluster Labels	1	2
Number of samples	11	13

'MIRSEQ CNMF' versus 'Time to Death'

P value = 0.0458 (logrank test), Q value = 1

Table S17. Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	24	4	0.2 - 211.2 (28.6)
subtype1	11	1	4.3 - 211.2 (31.1)
subtype2	13	3	0.2 - 106.1 (22.3)

Figure S13. Get High-res Image Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #1: 'Time to Death'

'MIRSEQ CNMF' versus 'AGE'

P value = 0.954 (Wilcoxon-test), Q value = 1

Table S18. Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	24	56.2 (13.1)
subtype1	11	57.3 (11.6)
subtype2	13	55.4 (14.7)

Figure S14. Get High-res Image Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #2: 'AGE'

'MIRSEQ CNMF' versus 'GENDER'

P value = 0.682 (Fisher's exact test), Q value = 1

Table S19. Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	13	11
subtype1	5	6
subtype2	8	5

Figure S15. Get High-res Image Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #3: 'GENDER'

'MIRSEQ CNMF' versus 'RACE'

P value = 0.825 (Fisher's exact test), Q value = 1

Table S20. Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #4: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	10	1	13
subtype1	4	0	7
subtype2	6	1	6

Figure S16. Get High-res Image Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #4: 'RACE'

Clustering Approach #5: 'MIRSEQ CHIERARCHICAL'

Table S21. Description of clustering approach #5: 'MIRSEQ CHIERARCHICAL'

Cluster Labels	1	2	3	4	5	6	7	8
Number of samples	2	2	3	4	5	2	3	3

'MIRSEQ CHIERARCHICAL' versus 'Time to Death'

P value = 0.443 (logrank test), Q value = 1

Table S22. Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	18	4	0.2 - 211.2 (33.6)
subtype3	3	0	15.0 - 196.6 (46.3)
subtype4	4	0	0.2 - 29.9 (11.3)
subtype5	5	1	4.6 - 74.1 (31.7)
subtype7	3	1	4.1 - 211.2 (52.0)
subtype8	3	2	41.2 - 106.1 (47.4)

Figure S17. Get High-res Image Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #1: 'Time to Death'

'MIRSEQ CHIERARCHICAL' versus 'AGE'

P value = 0.939 (Kruskal-Wallis (anova)), Q value = 1

Table S23. Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	18	55.9 (13.7)
subtype3	3	57.7 (15.3)
subtype4	4	61.2 (8.2)
subtype5	5	52.2 (19.4)
subtype7	3	55.0 (15.7)
subtype8	3	54.0 (12.1)

Figure S18. Get High-res Image Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #2: 'AGE'

'MIRSEQ CHIERARCHICAL' versus 'GENDER'

P value = 0.623 (Fisher's exact test), Q value = 1

Table S24. Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	10	8
subtype3	1	2
subtype4	3	1
subtype5	4	1
subtype7	1	2
subtype8	1	2

Figure S19. Get High-res Image Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #3: 'GENDER'

'MIRSEQ CHIERARCHICAL' versus 'RACE'

P value = 0.146 (Fisher's exact test), Q value = 1

Table S25. Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #4: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	8	1	9
subtype3	1	0	2
subtype4	4	0	0
subtype5	2	0	3
subtype7	1	0	2
subtype8	0	1	2

Figure S20. Get High-res Image Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #4: 'RACE'

Clustering Approach #6: 'MIRseq Mature CNMF subtypes'

Table S26. Description of clustering approach #6: 'MIRseq Mature CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	6	9	9

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

P value = 0.434 (logrank test), Q value = 1

Table S27. Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	24	4	0.2 - 211.2 (28.6)
subtype1	6	0	4.3 - 196.6 (27.0)
subtype2	9	2	0.2 - 106.1 (24.7)
subtype3	9	2	4.1 - 211.2 (31.7)

Figure S21. Get High-res Image Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'MIRseq Mature CNMF subtypes' versus 'AGE'

P value = 0.892 (Kruskal-Wallis (anova)), Q value = 1

Table S28. Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	24	56.2 (13.1)
subtype1	6	57.8 (14.2)
subtype2	9	57.8 (10.1)
subtype3	9	53.7 (15.9)

Figure S22. Get High-res Image Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #2: 'AGE'

'MIRseq Mature CNMF subtypes' versus 'GENDER'

P value = 0.512 (Fisher's exact test), Q value = 1

Table S29. Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	13	11
subtype1	2	4
subtype2	5	4
subtype3	6	3

Figure S23. Get High-res Image Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #3: 'GENDER'

'MIRseq Mature CNMF subtypes' versus 'RACE'

P value = 0.907 (Fisher's exact test), Q value = 1

Table S30. Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #4: 'RACE'

nPatients	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	10	1	13
subtype1	2	0	4
subtype2	4	1	4
subtype3	4	0	5

Figure S24. Get High-res Image Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #4: 'RACE'

Clustering Approach #7: 'MIRseq Mature cHierClus subtypes'

Table S31. Description of clustering approach #7: 'MIRseq Mature cHierClus subtypes'

Cluster Labels	1	2	3	4	5	6	7	8	9
Number of samples	2	3	3	2	3	2	4	3	2

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

P value = 0.428 (logrank test), Q value = 1

Table S32. Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	16	3	0.2 - 211.2 (28.2)
subtype2	3	0	0.2 - 24.7 (4.3)
subtype3	3	0	15.0 - 196.6 (46.3)
subtype5	3	1	22.3 - 106.1 (35.5)
subtype7	4	1	4.6 - 74.1 (25.6)
subtype8	3	1	4.1 - 211.2 (52.0)

Figure S25. Get High-res Image Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'MIRseq Mature cHierClus subtypes' versus 'AGE'

P value = 0.861 (Kruskal-Wallis (anova)), Q value = 1

Table S33. Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	16	57.0 (14.6)
subtype2	3	62.0 (14.2)
subtype3	3	57.7 (15.3)
subtype5	3	51.7 (20.8)
subtype7	4	58.2 (16.0)
subtype8	3	55.0 (15.7)

Figure S26. Get High-res Image Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'MIRseq Mature cHierClus subtypes' versus 'GENDER'

P value = 0.769 (Fisher's exact test), Q value = 1

Table S34. Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	8	8
subtype2	1	2
subtype3	1	2
subtype5	2	1
subtype7	3	1
subtype8	1	2

Figure S27. Get High-res Image Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

'MIRseq Mature cHierClus subtypes' versus 'RACE'

P value = 0.939 (Fisher's exact test), Q value = 1

Table S35. Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #4: 'RACE'

nPatients	ASIAN	WHITE
ALL	6	10
subtype2	1	2
subtype3	1	2
subtype5	2	1
subtype7	1	3
subtype8	1	2

Figure S28. Get High-res Image Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #4: 'RACE'

Methods & Data

Input

Cluster data file = DLBC-TP.mergedcluster.txt
Clinical data file = DLBC-TP.merged_data.txt
Number of patients = 33
Number of clustering approaches = 7
Number of selected clinical features = 4
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)

[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