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 5 clinical features across 10 patients, 2 significant findings 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 correlate to 'AGE'.
3 subtypes identified in current cancer cohort by 'MIRSEQ CHIERARCHICAL'. These subtypes do not correlate to any clinical features.
2 subtypes identified in current cancer cohort by 'MIRseq Mature CNMF subtypes'. These subtypes correlate to 'AGE'.
3 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 5 clinical features. Shown in the table are P values (Q values). Thresholded by P value < 0.05 and Q value < 0.25, 2 significant findings detected.


Clinical Features	Time to Death	AGE	NEOPLASM DISEASESTAGE	PATHOLOGY T STAGE	GENDER
Statistical Tests	logrank test	ANOVA	Chi-square test	NULL	Fisher's exact test
METHLYATION CNMF	100 (1.00)	0.8 (1.00)	1 (1.00)		1 (1.00)
RNAseq CNMF subtypes	0.0746 (1.00)	0.164 (1.00)	0.0302 (0.725)		0.486 (1.00)
RNAseq cHierClus subtypes	0.214 (1.00)	0.456 (1.00)	0.0838 (1.00)		0.286 (1.00)
MIRSEQ CNMF	100 (1.00)	0.00866 (0.242)	0.0302 (0.725)		0.486 (1.00)
MIRSEQ CHIERARCHICAL	0.802 (1.00)	0.00987 (0.257)	0.233 (1.00)		0.571 (1.00)
MIRseq Mature CNMF subtypes	100 (1.00)	0.00866 (0.242)	0.0302 (0.725)		0.486 (1.00)
MIRseq Mature cHierClus subtypes	0.802 (1.00)	0.00987 (0.257)	0.233 (1.00)		0.571 (1.00)

Clustering Approach #1: 'METHLYATION CNMF'

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

Cluster Labels	1	2	3	4
Number of samples	2	1	4	3

'METHLYATION CNMF' versus 'Time to Death'

P value = 100 (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	7	2	10.2 - 121.2 (29.2)
subtype3	4	1	10.2 - 57.5 (35.0)
subtype4	3	1	18.1 - 121.2 (29.2)

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

'METHLYATION CNMF' versus 'AGE'

P value = 0.8 (t-test), Q value = 1

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

	nPatients	Mean (Std.Dev)
ALL	7	42.9 (15.5)
subtype3	4	44.2 (18.8)
subtype4	3	41.0 (13.5)

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

'METHLYATION CNMF' versus 'NEOPLASM.DISEASESTAGE'

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

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

nPatients	STAGE I	STAGE IV
ALL	3	3
subtype3	1	2
subtype4	2	1

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

Clustering Approach #2: 'RNAseq CNMF subtypes'

Table S5. Description of clustering approach #2: 'RNAseq CNMF subtypes'

Cluster Labels	1	2	3	4
Number of samples	4	2	1	3

'RNAseq CNMF subtypes' versus 'Time to Death'

P value = 0.0746 (logrank test), Q value = 1

Table S6. Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	7	3	11.3 - 121.2 (29.2)
subtype1	4	3	11.3 - 37.1 (18.1)
subtype4	3	0	29.2 - 121.2 (46.8)

Figure S4. Get High-res Image Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq CNMF subtypes' versus 'AGE'

P value = 0.164 (t-test), Q value = 1

Table S7. Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	7	41.0 (16.4)
subtype1	4	48.5 (16.1)
subtype4	3	31.0 (12.3)

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

'RNAseq CNMF subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.0302 (Chi-square test), Q value = 0.72

Table S8. Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE IV
ALL	3	2	2
subtype1	0	2	2
subtype4	3	0	0

Figure S6. Get High-res Image Clustering Approach #2: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

Clustering Approach #3: 'RNAseq cHierClus subtypes'

Table S9. Description of clustering approach #3: 'RNAseq cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	3	4	3

'RNAseq cHierClus subtypes' versus 'Time to Death'

P value = 0.214 (logrank test), Q value = 1

Table S10. Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	10	4	10.2 - 121.2 (26.3)
subtype1	3	2	11.3 - 37.1 (18.1)
subtype2	4	1	18.1 - 121.2 (38.0)
subtype3	3	1	10.2 - 57.5 (23.3)

Figure S7. Get High-res Image Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq cHierClus subtypes' versus 'AGE'

P value = 0.456 (ANOVA), Q value = 1

Table S11. Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	10	44.2 (15.8)
subtype1	3	47.3 (19.5)
subtype2	4	36.2 (14.5)
subtype3	3	51.7 (14.0)

Figure S8. Get High-res Image Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'RNAseq cHierClus subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.0838 (Chi-square test), Q value = 1

Table S12. Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE IV
ALL	3	2	4
subtype1	0	2	1
subtype2	3	0	1
subtype3	0	0	2

Figure S9. Get High-res Image Clustering Approach #3: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

Clustering Approach #4: 'MIRSEQ CNMF'

Table S13. Description of clustering approach #4: 'MIRSEQ CNMF'

Cluster Labels	1	2	3
Number of samples	4	2	4

'MIRSEQ CNMF' versus 'Time to Death'

P value = 100 (logrank test), Q value = 1

Table S14. Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	8	2	10.2 - 121.2 (33.2)
subtype1	4	1	10.2 - 37.1 (17.3)
subtype3	4	1	29.2 - 121.2 (52.2)

Figure S10. Get High-res Image Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #1: 'Time to Death'

'MIRSEQ CNMF' versus 'AGE'

P value = 0.00866 (t-test), Q value = 0.24

Table S15. Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	8	45.6 (16.0)
subtype1	4	58.8 (5.3)
subtype3	4	32.5 (10.5)

Figure S11. Get High-res Image Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #2: 'AGE'

'MIRSEQ CNMF' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.0302 (Chi-square test), Q value = 0.72

Table S16. Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE IV
ALL	3	2	2
subtype1	0	2	2
subtype3	3	0	0

Figure S12. Get High-res Image Clustering Approach #4: 'MIRSEQ CNMF' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

Clustering Approach #5: 'MIRSEQ CHIERARCHICAL'

Table S17. Description of clustering approach #5: 'MIRSEQ CHIERARCHICAL'

Cluster Labels	1	2	3
Number of samples	3	3	4

'MIRSEQ CHIERARCHICAL' versus 'Time to Death'

P value = 0.802 (logrank test), Q value = 1

Table S18. Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	10	4	10.2 - 121.2 (26.3)
subtype1	3	2	18.1 - 121.2 (18.1)
subtype2	3	1	29.2 - 57.5 (46.8)
subtype3	4	1	10.2 - 37.1 (17.3)

Figure S13. Get High-res Image Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #1: 'Time to Death'

'MIRSEQ CHIERARCHICAL' versus 'AGE'

P value = 0.00987 (ANOVA), Q value = 0.26

Table S19. Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	10	44.2 (15.8)
subtype1	3	40.7 (14.0)
subtype2	3	28.3 (7.8)
subtype3	4	58.8 (5.3)

Figure S14. Get High-res Image Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #2: 'AGE'

'MIRSEQ CHIERARCHICAL' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.233 (Chi-square test), Q value = 1

Table S20. Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE IV
ALL	3	2	4
subtype1	1	0	2
subtype2	2	0	0
subtype3	0	2	2

Figure S15. Get High-res Image Clustering Approach #5: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

Clustering Approach #6: 'MIRseq Mature CNMF subtypes'

Table S21. Description of clustering approach #6: 'MIRseq Mature CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	4	2	4

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

P value = 100 (logrank test), Q value = 1

Table S22. Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	8	2	10.2 - 121.2 (33.2)
subtype1	4	1	10.2 - 37.1 (17.3)
subtype3	4	1	29.2 - 121.2 (52.2)

Figure S16. 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 = 0.00866 (t-test), Q value = 0.24

Table S23. Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	8	45.6 (16.0)
subtype1	4	58.8 (5.3)
subtype3	4	32.5 (10.5)

Figure S17. Get High-res Image Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #2: 'AGE'

'MIRseq Mature CNMF subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.0302 (Chi-square test), Q value = 0.72

Table S24. Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE IV
ALL	3	2	2
subtype1	0	2	2
subtype3	3	0	0

Figure S18. Get High-res Image Clustering Approach #6: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

Clustering Approach #7: 'MIRseq Mature cHierClus subtypes'

Table S25. Description of clustering approach #7: 'MIRseq Mature cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	4	3	3

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

P value = 0.802 (logrank test), Q value = 1

Table S26. Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	10	4	10.2 - 121.2 (26.3)
subtype1	4	1	10.2 - 37.1 (17.3)
subtype2	3	1	29.2 - 57.5 (46.8)
subtype3	3	2	18.1 - 121.2 (18.1)

Figure S19. 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 = 0.00987 (ANOVA), Q value = 0.26

Table S27. Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	10	44.2 (15.8)
subtype1	4	58.8 (5.3)
subtype2	3	28.3 (7.8)
subtype3	3	40.7 (14.0)

Figure S20. Get High-res Image Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'MIRseq Mature cHierClus subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.233 (Chi-square test), Q value = 1

Table S28. Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE IV
ALL	3	2	4
subtype1	0	2	2
subtype2	2	0	0
subtype3	1	0	2

Figure S21. Get High-res Image Clustering Approach #7: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

Methods & Data

Input

Cluster data file = ACC-TP.mergedcluster.txt
Clinical data file = ACC-TP.merged_data.txt
Number of patients = 10
Number of clustering approaches = 7
Number of selected clinical features = 5
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

Student's t-test analysis

For continuous numerical clinical features, two-tailed Student's t test with unequal variance (Lehmann and Romano 2005) was applied to compare the clinical values between two tumor subtypes using 't.test' function in R

Fisher's exact test

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

Chi-square test

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

ANOVA analysis

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

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] Lehmann and Romano, Testing Statistical Hypotheses (3E ed.), New York: Springer. ISBN 0387988645 (2005)

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

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

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

[8] 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