Skip to main content

Table 2 Results of the linear regression analyses on the global expression matrix calculated for the 1273 samples from 7 datasets (GSEs) combined using 5 different preprocessing and normalization methods

From: Survival marker genes of colorectal cancer derived from consistent transcriptomic profiling

FACTORS considered

Estimated coefficients

std. error

t value

p.value

Factor effect

(A) RMA

Intercept

6.925

0.014

512.610

<2e-16

(GSE14333+) GSE17536

0.387

0.019

20.230

<2e-16

yes

GSE31595

−1.212

0.019

−63.440

<2e-16

yes

GSE33113

−0.577

0.019

−30.210

<2e-16

yes

GSE38832

−0.355

0.019

−18.570

<2e-16

yes

GSE39084

−0.978

0.019

−51.180

<2e-16

yes

GSE39582

−1.375

0.019

−71.970

<2e-16

yes

(B) RMA plus Combat

Intercept

6.219

0.013

473.582

<2e-16

(GSE14333+) GSE17536

0.000

0.019

0.001

0.999

no

GSE31595

0.002

0.019

0.122

0.903

no

GSE33113

0.001

0.019

0.051

0.959

no

GSE38832

−0.001

0.019

−0.033

0.973

no

GSE39084

0.002

0.019

0.092

0.927

no

GSE39582

0.001

0.019

0.029

0.977

no

(C) fRMA

Intercept

6.535

0.015

450.434

<2e-16

(GSE14333+) GSE17536

−0.011

0.021

−0.553

0.580

no so much

GSE31595

0.089

0.021

4.329

0.000

yes

GSE33113

0.071

0.021

3.455

0.001

yes

GSE38832

0.054

0.021

2.641

0.008

yes

GSE39084

0.096

0.021

4.695

0.000

yes

GSE39582

0.089

0.021

4.336

0.000

yes

(D) fRMA plus Combat

Intercept

6.590

0.014

457.338

<2e-16

(GSE14333+) GSE17536

0.000

0.020

0.001

1.000

no

GSE31595

0.002

0.020

0.093

0.926

no

GSE33113

0.001

0.020

0.072

0.942

no

GSE38832

0.000

0.020

0.019

0.985

no

GSE39084

0.002

0.020

0.089

0.929

no

GSE39582

0.000

0.020

0.007

0.994

no

(E) fRMA plus mean centered

Intercept

0.000

0.000

−1.638

0.101

(GSE14333+) GSE17536

0.000

0.000

1.264

0.206

yes

GSE31595

0.000

0.000

0.288

0.773

no so much

GSE33113

0.000

0.000

1.605

0.108

yes

GSE38832

0.000

0.000

1.449

0.147

yes

GSE39084

0.000

0.000

−0.076

0.940

no

GSE39582

0.000

0.000

1.395

0.163

yes

  1. The methods applied were: (A) RMA; (B) RMA plus ComBat; (C) fRMA; (D) fRMA plus ComBat; (E) fRMA plus scaling of the data using mean-centered expression values. The linear regression is done to evaluate the “batch effect” (i.e. considering that the tested factors are the fact of “belonging” to a given dataset). Thus, when the p-value of the factors are significant (< 0.05), the “batch effect” remains on the overall expression signal. A marginal low significance was considered when p-values were < 0.20 in the case E