Table 3.

Overrepresented KEGG pathways among the top results from CCLS GWAS of childhood ALL among Hispanics and identification of important genes using a causal inference approach

KEGG pathwayGenesObsExpRatioPPadjustORa (95% CI)Important genesb
Axon guidanceLRRC4C, PLXNC1, SLIT3, EPHB1, NTN1, UNC5B, GNAI1, NGEF80.5514.469.64 × 10−85.1 × 10−61.59 (1.35–1.88)UNC5B
EPHB1
PLXNC1
Protein digestion and absorptionCPA2, COL4A2, SLC7A8, COL5A1, COL6A650.3514.392.71 × 10−57.1 × 10−42.03 (1.60–2.58)CPA2
COL6A6
COL5A1
SLC7A8
MelanogenesisTCF7L1, CAMK2D, DVL3, MAP2K2, GNAI150.4311.547.81 × 10−50.00141.58 (1.35–1.84)DVL3
TCF7L1
CAMK2D
MAP2K2
Leukocyte transendothelial migrationITGAL, VAV3, MYL2, CTNNA2, GNAI150.5010.052.1 × 10−40.00211.64 (1.35–1.98)VAV3
CTNNA2
Focal adhesionVAV3, MYL2, COL4A2, TLN1, COL5A1, COL6A660.866.992.1 × 10−40.00211.96 (1.60–2.41)VAV3
COL6A6
COL5A1
Endometrial cancerTCF7L1, MAP2K2, CTNNA230.2213.450.0010.0131.53 (1.25–1.87)TCF7L1
MAP2K2
CTNNA2
GliomaCAMK2D, MAP2K2, CDK630.2810.80.0020.0141.59 (1.33–1.91)CDK6
CAMK2D
MAP2K2
Pathways in cancerTCF7L1, DVL3, COL4A2, MAP2K2, CDK6, CTNNA261.404.290.0030.0141.63 (1.41–1.90)DVL3
CDK6
TCF7L1
MAP2K2
CTNNA2
TCF7L1
Tight junctionMYL2, RRAS2, CTNNA2, GNAI140.577.060.0020.0141.52 (1.24–1.87)RRAS2
CTNNA2
Regulation of actin cytoskeletonITGAL, VAV3, MYL2, RRAS2, MAP2K250.915.470.0020.0141.69 (1.44–2.00)VAV3
MAP2K2
RRAS2

NOTE: SNPs which showed association with childhood ALL (P < 0.001) were included in this study. Of the 187 genes submitted for analysis, 185 were incorporated for analysis using a hypergeometric test to compare the submitted list to a reference of all human genes using WebGestalt v.2 (http://bioinfo.vanderbilt.edu/webgestalt/). The analysis was limited to KEGG pathways where at least two genes were present in the submitted list. The top 10 ranking KEGG pathways are shown. Adjusted P values were based on correlation for FDR using BH procedure.

Abbreviations: Exp, expected; KEGG, Kyoto Encyclopedia of Genes and Genomes; Obs, observed.

  • a OR and 95% CI for cumulative effects of SNPs within each pathway on the risk of childhood ALL was calculated using a logistic regression model.

  • b Genes were selected on the basis of the P values from TMLE (P < 0.05).