Correlation between mRNAseq expression and clinical features

Ovarian Serous Cystadenocarcinoma (Primary solid tumor)

16 April 2014 | analyses__2014_04_16

Maintainer Information

Citation Information

Maintained by Juok Cho (Broad Institute)

Cite as Broad Institute TCGA Genome Data Analysis Center (2014): Correlation between mRNAseq expression and clinical features. Broad Institute of MIT and Harvard. doi:10.7908/C11J98D0

Overview

Introduction

This pipeline uses various statistical tests to identify mRNAs whose expression levels correlated to selected clinical features.

Summary

Testing the association between 18555 genes and 3 clinical features across 261 samples, statistically thresholded by Q value < 0.05, 2 clinical features related to at least one genes.

1 gene correlated to 'Time to Death'.

KRTCAP3|200634

57 genes correlated to 'AGE'.

STS|412 , LRP6|4040 , C8ORF55|51337 , EIF4E3|317649 , C12ORF4|57102 , ...

No genes correlated to 'KARNOFSKY.PERFORMANCE.SCORE'

Results

Overview of the results

Complete statistical result table is provided in Supplement Table 1

Table 1. Get Full Table This table shows the clinical features, statistical methods used, and the number of genes that are significantly associated with each clinical feature at Q value < 0.05.

Clinical feature	Statistical test	Significant genes	Associated with		Associated with
Time to Death	Cox regression test	N=1	shorter survival	N=0	longer survival	N=1
AGE	Spearman correlation test	N=57	older	N=30	younger	N=27
KARNOFSKY PERFORMANCE SCORE	Spearman correlation test	N=0

Clinical variable #1: 'Time to Death'

One gene related to 'Time to Death'.

Table S1. Basic characteristics of clinical feature: 'Time to Death'


Time to Death	Duration (Months)	0.3-180.2 (median=28.2)
	censored	N = 112
	death	N = 147

	Significant markers	N = 1
	associated with shorter survival	0
	associated with longer survival	1

List of one gene significantly associated with 'Time to Death' by Cox regression test

Table S2. Get Full Table List of one gene significantly associated with 'Time to Death' by Cox regression test

	HazardRatio	Wald_P	Q	C_index
KRTCAP3\|200634	0.68	2.131e-06	0.04	0.403

Figure S1. Get High-res Image As an example, this figure shows the association of KRTCAP3|200634 to 'Time to Death'. four curves present the cumulative survival rates of 4 quartile subsets of patients. P value = 2.13e-06 with univariate Cox regression analysis using continuous log-2 expression values.

Clinical variable #2: 'AGE'

57 genes related to 'AGE'.

Table S3. Basic characteristics of clinical feature: 'AGE'


AGE	Mean (SD)	58.96 (11)
	Significant markers	N = 57
	pos. correlated	30
	neg. correlated	27

List of top 10 genes significantly correlated to 'AGE' by Spearman correlation test

Table S4. Get Full Table List of top 10 genes significantly correlated to 'AGE' by Spearman correlation test

	SpearmanCorr	corrP	Q
STS\|412	-0.3934	7.899e-11	1.47e-06
LRP6\|4040	0.3678	1.483e-09	2.75e-05
C8ORF55\|51337	-0.3588	3.909e-09	7.25e-05
EIF4E3\|317649	-0.3555	5.541e-09	0.000103
C12ORF4\|57102	0.3551	5.776e-09	0.000107
APPL2\|55198	0.3478	1.242e-08	0.00023
PDHA1\|5160	-0.3456	1.551e-08	0.000288
GREB1\|9687	-0.3446	1.716e-08	0.000318
CLSTN3\|9746	0.3431	1.993e-08	0.00037
ADAM15\|8751	-0.3411	2.442e-08	0.000453

Figure S2. Get High-res Image As an example, this figure shows the association of STS|412 to 'AGE'. P value = 7.9e-11 with Spearman correlation analysis. The straight line presents the best linear regression.

Clinical variable #3: 'KARNOFSKY.PERFORMANCE.SCORE'

No gene related to 'KARNOFSKY.PERFORMANCE.SCORE'.

Table S5. Basic characteristics of clinical feature: 'KARNOFSKY.PERFORMANCE.SCORE'


KARNOFSKY.PERFORMANCE.SCORE	Mean (SD)	74.29 (12)
	Score	N
	60	5
	80	8
	100	1

	Significant markers	N = 0

Methods & Data

Input

Expresson data file = OV-TP.uncv2.mRNAseq_RSEM_normalized_log2.txt
Clinical data file = OV-TP.merged_data.txt
Number of patients = 261
Number of genes = 18555
Number of clinical features = 3

Survival analysis

For survival clinical features, Wald's test in univariate Cox regression analysis with proportional hazards model (Andersen and Gill 1982) was used to estimate the P values using the 'coxph' function in R. Kaplan-Meier survival curves were plot using the four quartile subgroups of patients based on expression levels

Correlation analysis

For continuous numerical clinical features, Spearman's rank correlation coefficients (Spearman 1904) and two-tailed P values were estimated using 'cor.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] Andersen and Gill, Cox's regression model for counting processes, a large sample study, Annals of Statistics 10(4):1100-1120 (1982)

[2] Spearman, C, The proof and measurement of association between two things, Amer. J. Psychol 15:72-101 (1904)

[3] 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