Rare variant collapsing in conjunction with mean log p-value and gradient boosting approaches applied to Genetic Analysis Workshop 17 data

Table 1 Major IPA pathways identified by the MLP approach using five statistics for rare and common variants and their corresponding p-values

Statistic	Top 6 pathways selected	p-value
Rare variants
Gene-wise collapsing	1. Androgen and estrogen metabolism	0.0006
	2. Sphingolipid metabolism	0.0006
	3. Phenylalanine metabolism	0.0015
	4. Death receptor signaling	0.002
	5. Stilbene, coumarine, and lignin biosynthesis	0.0036
	6. TWEAK signaling	0.0043
Minimum p-value of 5-kb sliding window collapsing within a gene	1. Notch signaling	0.0248
	2. Hypoxia signaling in the cardiovascular system	0.0403
	3. Nitric oxide signaling in the cardiovascular system	0.0409
	4. VEGF signaling	0.0585
	5. Glutamate receptor signaling	0.0642
	6. Glutamate metabolism	0.0741
Mean log p-value of 5-kb sliding window collapsing within a gene	1. Cyanoamino acid metabolism	0.0032
	2. Ubiquinone biosynthesis	0.0148
	3. Nitrogen metabolism	0.0267
	4. Alanine and aspartate metabolism	0.0392
	5. GABA receptor signaling	0.0423
	6. FXR/RXR activation	0.0438
Common variants
Minimum p-value among SNPs	1. Apoptosis signaling	0.0083
	2. Pyrimidine metabolism	0.0232
	3. CNTF signaling	0.0429
	4. FLT3 signaling in hematopoietic progenitor cells	0.0601
	5. Role of NANOG in mammalian embryonic stem cell pluripotency	0.0847
	6. EGF signaling	0.0908
Mean log p-value for SNPs in a gene	1. Pyrimidine metabolism	0.0021
	2. CNTF signaling	0.0127
	3. Melanocyte development and pigmentation signaling	0.0221
	4. JAK/Stat signaling	0.0327
	5. IL-15 signaling	0.0356
	6. FLT3 signaling in hematopoietic progenitor cells	0.045

ISSN: 1753-6561