Correlation between gene methylation status and clinical features
Acute Myeloid Leukemia (Primary blood derived cancer - Peripheral blood)
28 January 2016  |  analyses__2016_01_28
Maintainer Information
Citation Information
Maintained by Juok Cho (Broad Institute)
Cite as Broad Institute TCGA Genome Data Analysis Center (2016): Correlation between gene methylation status and clinical features. Broad Institute of MIT and Harvard. doi:10.7908/C1T72GVP
Overview
Introduction

This pipeline uses various statistical tests to identify genes whose promoter methylation levels correlated to selected clinical features. The input file "LAML-TB.meth.by_min_clin_corr.data.txt" is generated in the pipeline Methylation_Preprocess in stddata run.

Summary

Testing the association between 11414 genes and 4 clinical features across 194 samples, statistically thresholded by P value < 0.05 and Q value < 0.3, 4 clinical features related to at least one genes.

  • 30 genes correlated to 'DAYS_TO_DEATH_OR_LAST_FUP'.

    • RLN2 ,  TLN1 ,  DEXI ,  FRMD5 ,  ZFYVE21 ,  ...

  • 30 genes correlated to 'YEARS_TO_BIRTH'.

    • TMEM20 ,  IGFBP7 ,  BTBD3 ,  CAMK4 ,  KHNYN ,  ...

  • 14 genes correlated to 'GENDER'.

    • DECR1 ,  NRXN2 ,  KIAA0226 ,  TMSL3 ,  VAMP2 ,  ...

  • 2 genes correlated to 'RACE'.

    • TNFRSF21 ,  TNNI3

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 P value < 0.05 and Q value < 0.3.

Clinical feature Statistical test Significant genes Associated with                 Associated with
DAYS_TO_DEATH_OR_LAST_FUP Cox regression test N=30   N=NA   N=NA
YEARS_TO_BIRTH Spearman correlation test N=30 older N=7 younger N=23
GENDER Wilcoxon test N=14 male N=14 female N=0
RACE Kruskal-Wallis test N=2        
Clinical variable #1: 'DAYS_TO_DEATH_OR_LAST_FUP'

30 genes related to 'DAYS_TO_DEATH_OR_LAST_FUP'.

Table S1.  Basic characteristics of clinical feature: 'DAYS_TO_DEATH_OR_LAST_FUP'

DAYS_TO_DEATH_OR_LAST_FUP Duration (Months) 0-94.1 (median=12)
  censored N = 65
  death N = 128
     
  Significant markers N = 30
  associated with shorter survival NA
  associated with longer survival NA
List of top 10 genes differentially expressed by 'DAYS_TO_DEATH_OR_LAST_FUP'

Table S2.  Get Full Table List of top 10 genes significantly associated with 'Time to Death' by Cox regression test. For the survival curves, it compared quantile intervals at c(0, 0.25, 0.50, 0.75, 1) and did not try survival analysis if there is only one interval.

logrank_P Q C_index
RLN2 9.21e-07 0.011 0.406
TLN1 1.99e-06 0.011 0.338
DEXI 1.26e-05 0.02 0.42
FRMD5 1.29e-05 0.02 0.387
ZFYVE21 1.31e-05 0.02 0.34
PARP3 1.35e-05 0.02 0.383
EREG 1.4e-05 0.02 0.398
MPO 1.54e-05 0.02 0.61
SPINK2 1.59e-05 0.02 0.388
VPS26B 2.05e-05 0.023 0.365
Clinical variable #2: 'YEARS_TO_BIRTH'

30 genes related to 'YEARS_TO_BIRTH'.

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

YEARS_TO_BIRTH Mean (SD) 55.11 (16)
  Significant markers N = 30
  pos. correlated 7
  neg. correlated 23
List of top 10 genes differentially expressed by 'YEARS_TO_BIRTH'

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

SpearmanCorr corrP Q
TMEM20 -0.4547 2.74e-11 3.13e-07
IGFBP7 0.4357 2.155e-10 1.23e-06
BTBD3 0.4237 7.463e-10 2.84e-06
CAMK4 -0.3929 1.46e-08 3.88e-05
KHNYN -0.3902 1.857e-08 3.88e-05
JAKMIP1 -0.3892 2.037e-08 3.88e-05
IRS2 0.3831 3.507e-08 5.72e-05
SASH1 -0.3671 1.405e-07 2e-04
FRMD5 -0.3646 1.723e-07 0.000218
HIST3H2BB -0.362 2.139e-07 0.000229
Clinical variable #3: 'GENDER'

14 genes related to 'GENDER'.

Table S5.  Basic characteristics of clinical feature: 'GENDER'

GENDER Labels N
  FEMALE 89
  MALE 105
     
  Significant markers N = 14
  Higher in MALE 14
  Higher in FEMALE 0
List of top 10 genes differentially expressed by 'GENDER'

Table S6.  Get Full Table List of top 10 genes differentially expressed by 'GENDER'. 0 significant gene(s) located in sex chromosomes is(are) filtered out.

W(pos if higher in 'MALE') wilcoxontestP Q AUC
DECR1 540 2.874e-26 3.28e-22 0.9422
NRXN2 1346 1.401e-17 8e-14 0.856
KIAA0226 1689 1.935e-14 7.36e-11 0.8193
TMSL3 2042 1.489e-11 3.46e-08 0.7815
VAMP2 2043 1.515e-11 3.46e-08 0.7814
CASP2 2163 1.207e-10 2.3e-07 0.7685
LRRC2 6936 6.352e-09 1.04e-05 0.7422
OXSR1 2652 2.176e-07 0.00031 0.7162
NDUFS1 2932 8.002e-06 0.0101 0.6862
ZNF395 3193 0.0001474 0.168 0.6583
Clinical variable #4: 'RACE'

2 genes related to 'RACE'.

Table S7.  Basic characteristics of clinical feature: 'RACE'

RACE Labels N
  ASIAN 2
  BLACK OR AFRICAN AMERICAN 13
  WHITE 177
     
  Significant markers N = 2
List of 2 genes differentially expressed by 'RACE'

Table S8.  Get Full Table List of 2 genes differentially expressed by 'RACE'

kruskal_wallis_P Q
TNFRSF21 6.333e-06 0.0473
TNNI3 8.289e-06 0.0473
Methods & Data
Input
  • Expresson data file = LAML-TB.meth.by_min_clin_corr.data.txt

  • Clinical data file = LAML-TB.merged_data.txt

  • Number of patients = 194

  • Number of genes = 11414

  • Number of clinical features = 4

Selected clinical features
  • Further details on clinical features selected for this analysis, please find a documentation on selected CDEs (Clinical Data Elements). The first column of the file is a formula to convert values and the second column is a clinical parameter name.

  • Survival time data

    • Survival time data is a combined value of days_to_death and days_to_last_followup. For each patient, it creates a combined value 'days_to_death_or_last_fup' using conversion process below.

      • if 'vital_status'==1(dead), 'days_to_last_followup' is always NA. Thus, uses 'days_to_death' value for 'days_to_death_or_fup'

      • if 'vital_status'==0(alive),

        • if 'days_to_death'==NA & 'days_to_last_followup'!=NA, uses 'days_to_last_followup' value for 'days_to_death_or_fup'

        • if 'days_to_death'!=NA, excludes this case in survival analysis and report the case.

      • if 'vital_status'==NA,excludes this case in survival analysis and report the case.

    • cf. In certain diesase types such as SKCM, days_to_death parameter is replaced with time_from_specimen_dx or time_from_specimen_procurement_to_death .

  • This analysis excluded clinical variables that has only NA values.

Survival analysis

For survival clinical features, logrank test in univariate Cox regression analysis with proportional hazards model (Andersen and Gill 1982) was used to estimate the P values comparing quantile intervals using the 'coxph' function in R. Kaplan-Meier survival curves were plotted using quantile intervals at c(0, 0.25, 0.50, 0.75, 1). If there is only one interval group, it will not try survival analysis.

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

Wilcoxon rank sum test (Mann-Whitney U test)

For two groups (mutant or wild-type) of continuous type of clinical data, wilcoxon rank sum test (Mann and Whitney, 1947) was applied to compare their mean difference using 'wilcox.test(continuous.clinical ~ as.factor(group), exact=FALSE)' function in R. This test is equivalent to the Mann-Whitney test.

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] Mann and Whitney, On a Test of Whether one of Two Random Variables is Stochastically Larger than the Other, Annals of Mathematical Statistics 18 (1), 50-60 (1947)
[4] 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)