This pipeline uses various statistical tests to identify mRNAs whose expression levels correlated to selected clinical features.
Testing the association between 12042 genes and 5 clinical features across 519 samples, statistically thresholded by Q value < 0.05, 4 clinical features related to at least one genes.

21 genes correlated to 'Time to Death'.

CLEC5A , EFEMP2 , NCOA4 , ATP5C1 , DIRAS3 , ...

76 genes correlated to 'AGE'.

RANBP17 , FBXO17 , TUSC3 , KIAA0495 , NOL3 , ...

23 genes correlated to 'GENDER'.

DDX3Y , RPS4Y1 , JARID1D , EIF1AY , NLGN4Y , ...

2 genes correlated to 'RADIATIONS.RADIATION.REGIMENINDICATION'.

NPAT , HOXD10

No genes correlated to 'KARNOFSKY.PERFORMANCE.SCORE'
Complete statistical result table is provided in Supplement Table 1
Clinical feature  Statistical test  Significant genes  Associated with  Associated with  

Time to Death  Cox regression test  N=21  shorter survival  N=11  longer survival  N=10 
AGE  Spearman correlation test  N=76  older  N=42  younger  N=34 
GENDER  t test  N=23  male  N=11  female  N=12 
KARNOFSKY PERFORMANCE SCORE  Spearman correlation test  N=0  
RADIATIONS RADIATION REGIMENINDICATION  t test  N=2  yes  N=1  no  N=1 
Time to Death  Duration (Months)  0.1127.6 (median=9.9) 
censored  N = 116  
death  N = 403  
Significant markers  N = 21  
associated with shorter survival  11  
associated with longer survival  10 
HazardRatio  Wald_P  Q  C_index  

CLEC5A  1.23  7.107e08  0.00086  0.584 
EFEMP2  1.3  7.708e08  0.00093  0.542 
NCOA4  0.56  8.196e08  0.00099  0.442 
ATP5C1  0.59  8.228e08  0.00099  0.451 
DIRAS3  1.22  1.111e07  0.0013  0.558 
RANBP17  0.46  1.833e07  0.0022  0.427 
ANKRD26  0.39  2.458e07  0.003  0.446 
HIST3H2A  0.82  3.552e07  0.0043  0.427 
ZIC3  0.48  6.625e07  0.008  0.444 
FZD7  1.23  1.054e06  0.013  0.556 
AGE  Mean (SD)  57.68 (14) 
Significant markers  N = 76  
pos. correlated  42  
neg. correlated  34 
SpearmanCorr  corrP  Q  

RANBP17  0.316  1.677e13  2.02e09 
FBXO17  0.3024  1.966e12  2.37e08 
TUSC3  0.2972  4.787e12  5.76e08 
KIAA0495  0.279  9.796e11  1.18e06 
NOL3  0.2745  2.002e10  2.41e06 
PPA1  0.2725  2.734e10  3.29e06 
H2AFY2  0.2638  1.037e09  1.25e05 
DRG2  0.2628  1.203e09  1.45e05 
NCOA4  0.2621  1.343e09  1.62e05 
ENOSF1  0.2585  2.273e09  2.73e05 
GENDER  Labels  N 
FEMALE  204  
MALE  315  
Significant markers  N = 23  
Higher in MALE  11  
Higher in FEMALE  12 
T(pos if higher in 'MALE')  ttestP  Q  AUC  

DDX3Y  37.55  8.529e142  1.03e137  0.96 
RPS4Y1  40.23  7.983e140  9.61e136  0.9521 
JARID1D  34.79  1.118e136  1.35e132  0.9603 
EIF1AY  34.88  6.512e134  7.84e130  0.9536 
NLGN4Y  30.85  4.165e117  5.01e113  0.9485 
USP9Y  21.13  1.174e71  1.41e67  0.917 
CYORF15B  19.36  3.851e63  4.64e59  0.9038 
UTY  19.74  4.402e60  5.3e56  0.8998 
ZFX  12.48  7.793e30  9.38e26  0.8205 
HDHD1A  12.39  1.389e29  1.67e25  0.8043 
No gene related to 'KARNOFSKY.PERFORMANCE.SCORE'.
KARNOFSKY.PERFORMANCE.SCORE  Mean (SD)  77.12 (14) 
Significant markers  N = 0 
2 genes related to 'RADIATIONS.RADIATION.REGIMENINDICATION'.
RADIATIONS.RADIATION.REGIMENINDICATION  Labels  N 
NO  348  
YES  171  
Significant markers  N = 2  
Higher in YES  1  
Higher in NO  1 

Expresson data file = GBMTP.medianexp.txt

Clinical data file = GBMTP.clin.merged.picked.txt

Number of patients = 519

Number of genes = 12042

Number of clinical features = 5
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. KaplanMeier survival curves were plot using the four quartile subgroups of patients based on expression levels
For continuous numerical clinical features, Spearman's rank correlation coefficients (Spearman 1904) and twotailed P values were estimated using 'cor.test' function in R
For twoclass clinical features, twotailed Student's t test with unequal variance (Lehmann and Romano 2005) was applied to compare the log2expression levels between the two clinical classes using 't.test' function in R
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.
This is an experimental feature. The full results of the analysis summarized in this report can be downloaded from the TCGA Data Coordination Center.