Skip to main content

Table 5 Random Forest results of the analysis of ACTG data with HDL-c levels as trait and SNPs from ApoCIII, ApoE, EL and HL genes and race/ethnicity as predictors

From: Application of two machine learning algorithms to genetic association studies in the presence of covariates

 

Include*

Ignore*

Residualize*

White

(n = 317)

Stratify* Black

(n = 92)

Hispanic

(n = 103)

ApoC-III

      

-482C/T (rs2854117)

12.68(0.63)

11.81(0.60)

13.07(0.66)

13.05(0.66)

-2.04(0.96)

1.19(0.97)

-455T/C (rs2854116)

9.04(0.66)

9.69(0.66)

10.13(0.67)

8.90(0.69)

-0.57(0.98)

1.27(0.98)

intron 1 (466)G/C (rs2070669)

4.70(0.93)

5.85(0.88)

5.91(0.90)

4.55(0.91)

-2.78(0.94)

-2.62(1.00)

Gly34Gly C/T (rs4520)

8.86(1.12)

5.51(1.03)

4.77(1.07)

2.99(1.02)

0.91(0.99)

11.87(0.91)

exon 4 SstI 4348(5) C/G(rs5128)

0.60(1.01)

2.12(1.04)

2.34(1.04)

2.91(1.03)

-3.20(0.97)

2.95(1.00)

ApoE

      

Arg112Cys T/C (rs429358)

4.87(1.08)

1.45(1.00)

2.29(1.06)

6.29(1.08)

-5.22(0.96)

6.30(0.98)

Arg158Cys T/C (rs7412)

6.02(0.98)

7.49(1.00)

7.49(0.93)

5.08(0.92)

-5.20(0.97)

2.69(0.98)

EL

      

rs12970066,

3.94(1.03)

4.00(0.97)

5.54(0.97)

1.12(0.97)

8.57(0.91)

-1.81(1.05)

Asn396Ser,

7.06(1.03)

8.30(1.05)

8.39(1.10)

2.27(1.04)

5.93(0.90)

2.09(0.97)

rs3829632 (-1309A/G)

-1.10 (0.99)

1.24(0.98)

2.25(1.04)

-1.64(0.98)

0.00(0.00)

-2.25(0.96)

HL

      

rs2070895

11.81(1.04)

7.86 (1.09)

5.54(0.99)

5.58(1.11)

-1.73(0.95)

2.82(0.97)

rs12595191

-0.93(0.97)

-1.77(1.00)

-1.07(0.99)

-3.62(1.00)

-1.99(0.99)

0.07(0.99)

rs690

10.41(1.08)

1.28(0.99)

0.53(0.98)

3.93(1.05)

-3.98(0.97)

9.61(0.96)

rs6084

7.42(1.01)

6.47(1.01)

6.27(1.06)

-1.15(1.01)

9.31(0.86)

10.97(0.90)

Race/ethnicity

21.58(1.15)

NA

NA

NA

NA

NA

  1. " * " indicates the approach to handling the race/ethnicity covariate;
  2. "NA" indicates that the predictor was not included in the analysis.
  3. The two highest importance scores from RF are in bold.