Genome-wide association study combined with biological context can reveal more disease-related SNPs altering microRNA target seed sites

Table 3 Statistics of the 12 GWAS datasets

Dataset	Disease	#Enriched GO-terms	#Disease-related GO-terms	Fisher-p	#Candidate SNPs	#PubMed SNPs	Fisher-p
phs000021.pha002857	Schizophrenia	48	9	0.0003	49	3	0.05
phs000089.pha002868	Parkinson’s disease	49	7	0.001	22	2	0.05
phs000100.pha002839	Type 2 diabetes	29	8	8 × 10⁻⁵	27	3	0.009
phs000122.pha002848	Systemic lupus erythematosus	70	21	4 × 10⁻²⁰	19	7	4 × 10⁻⁷
phs000124.pha002845	Neuroblastoma	92	5	0.003	28	4	0.001
phs000147.pha002853	Breast cancer	26	12	6 × 10⁻⁶	29	2	0.08
phs000171.pha002861	Multiple sclerosis	56	3	0.02	17	2	0.03
phs000182.pha002890	Age-related macular degeneration	42	0	1	0	0	1
phs000206.pha002889	Pancreatic cancer	27	13	3 × 10⁻⁹	51	3	0.05
phs000207.pha002878	Prostate cancer	46	26	7 × 10⁻¹⁴	36	2	0.1
phs000216.pha002867	Systemic lupus erythematosus	23	3	0.009	7	3	2 × 10⁻⁴
phs000423.pha003542	Coronary artery disease	19	4	2 × 10⁻⁴	17	0

‘#Enriched GO-terms’ denotes the number of enriched GO-terms for the host genes of MRESS SNPs with p-value ≤ 0.01 in each dataset. ‘#Disease-related GO-terms’ denotes the number of enriched GO-terms shared between the host genes and the disease-related genes collected from CTD, KEGG and Gene Card databases in each dataset. The probability of obtaining such number of disease-related GO-terms is calculated by fisher exact test. ‘#Candidate SNPs’ denotes the number of MRESS SNPs passing our selection standards in each dataset, and ‘PubMed SNPs’ denotes how many of them are reported to be deleterious by PubMed query, the probability of the observation is also calculated by fisher exact test. We can figure out from the table that most of the p-values are below the significant level of 0.05, which indicates MRESS SNPs as causal variants in most of the GWAS datasets.

ISSN: 1471-2164