Limits...
Allele-specific transcription factor binding to common and rare variants associated with disease and gene expression.

Cavalli M, Pan G, Nord H, Wallerman O, Wallén Arzt E, Berggren O, Elvers I, Eloranta ML, Rönnblom L, Lindblad Toh K, Wadelius C - Hum. Genet. (2016)

Bottom Line: We found 9962 candidate regulatory SNPs, of which 16 % were rare and showed evidence of larger functional effect than common ones.Functionally rare variants may explain divergent GWAS results between populations and are candidates for a partial explanation of the missing heritability.Furthermore, by examining GWAS loci we found >400 allele-specific candidate SNPs, 141 of which were highly relevant in our cell types.

View Article: PubMed Central - PubMed

Affiliation: Science for Life Laboratory, Department of Immunology, Genetics and Pathology, Uppsala University, Uppsala, Sweden.

ABSTRACT
Genome-wide association studies (GWAS) have identified a large number of disease-associated SNPs, but in few cases the functional variant and the gene it controls have been identified. To systematically identify candidate regulatory variants, we sequenced ENCODE cell lines and used public ChIP-seq data to look for transcription factors binding preferentially to one allele. We found 9962 candidate regulatory SNPs, of which 16 % were rare and showed evidence of larger functional effect than common ones. Functionally rare variants may explain divergent GWAS results between populations and are candidates for a partial explanation of the missing heritability. The majority of allele-specific variants (96 %) were specific to a cell type. Furthermore, by examining GWAS loci we found >400 allele-specific candidate SNPs, 141 of which were highly relevant in our cell types. Functionally validated SNPs support identification of an SNP in SYNGR1 which may expose to the risk of rheumatoid arthritis and primary biliary cirrhosis, as well as an SNP in the last intron of COG6 exposing to the risk of psoriasis. We propose that by repeating the ChIP-seq experiments of 20 selected transcription factors in three to ten people, the most common polymorphisms can be interrogated for allele-specific binding. Our strategy may help to remove the current bottleneck in functional annotation of the genome.

No MeSH data available.


Related in: MedlinePlus

AS-SNPs associated with GWAS-SNPs. aTop AS-SNPs associated with GWAS-SNPs intersecting the cell-specific collections of AS-SNPs with the full GWAS catalog. Numbers are reported for common and rare AS-SNPs that are direct hits in the GWAS catalog or in LD with GWAS-SNPs. Bottom AS-SNPs associated with GWAS-SNPs intersecting the cell-specific collections of AS-SNPs with GWAS-SNPs associated with cell-specific traits. b Model representation of the networks of interactions observed between AS-SNPs and GWAS-SNPs. The dotted red box highlights the simplest scenario with one AS-SNP in LD with a GWAS-SNP. The tables report the numbers of instances observed in each cell line where one AS-SNP is in LD with several GWAS-SNPs or one GWAS-SNPs is in LD with different AS-SNPs, or where GWAS-SNPs were also AS-SNPs. c Four AS-SNPs, located in three different regulatory elements, interact with several GWAS-SNPs associated with autoimmune diseases
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4835527&req=5

Fig3: AS-SNPs associated with GWAS-SNPs. aTop AS-SNPs associated with GWAS-SNPs intersecting the cell-specific collections of AS-SNPs with the full GWAS catalog. Numbers are reported for common and rare AS-SNPs that are direct hits in the GWAS catalog or in LD with GWAS-SNPs. Bottom AS-SNPs associated with GWAS-SNPs intersecting the cell-specific collections of AS-SNPs with GWAS-SNPs associated with cell-specific traits. b Model representation of the networks of interactions observed between AS-SNPs and GWAS-SNPs. The dotted red box highlights the simplest scenario with one AS-SNP in LD with a GWAS-SNP. The tables report the numbers of instances observed in each cell line where one AS-SNP is in LD with several GWAS-SNPs or one GWAS-SNPs is in LD with different AS-SNPs, or where GWAS-SNPs were also AS-SNPs. c Four AS-SNPs, located in three different regulatory elements, interact with several GWAS-SNPs associated with autoimmune diseases

Mentions: The GWAS catalog has entries for a wide spectrum of traits that are expressed in different cell types, all of which were not studied in this experiment. Molecular events that could be explained by the cells investigated are for example immune-mediated diseases for GM12878 and K562, and neurological diseases for SK-N-SH (Table S3 in Supplementary material 1). Consequently, we searched the catalog for these cell-specific traits. H1-hESC was not investigated. The SNP with the strongest association (GWAS top hit) was collected from the catalog and SNPs in high LD (r2 > 0.8) were identified and intersected with the collection of AS-SNPs. For 36 traits, we found 141 AS-SNPs. We investigated the number of candidate regulatory SNPs found when searching the GWAS catalog using matched random sets of non-AS-SNPs. We found significantly more AS-SNPs than expected by chance with a clear enrichment (5- to 13-fold) in the overlap of AS-SNPs as compared to random sets of non-AS-SNPs (Fig S3 in Supplementary material 1). Out of the 141 unique AS-SNPs that were candidates to explain GWAS signals, only 15 were the particular SNP reported in the GWAS catalog and the other 126 were in the high-LD interval and 10 of these were rare AS-SNPs (Table 2; Fig. 3a, b and Tables S8, S11, S13 in Supplementary material 3). We grouped the SNPs in 1 Mb loci, and in immune cells, GM12878 and K562, we found candidate functional SNPs at 71 loci and in SK-N-SH at 14 loci, giving a total of functional candidates at 85 loci. We compared our candidates with the total number of mapped regions for different diseases and detected tentative functional variants for 11 % of loci mapped for psoriasis, 9 % for SLE and self-reported allergy, 8 % for inflammatory bowel disease/Crohn’s disease/ulcerative colitis and 6 % for type 1 diabetes, suggesting new functional regulatory elements for many common diseases.Table 2


Allele-specific transcription factor binding to common and rare variants associated with disease and gene expression.

Cavalli M, Pan G, Nord H, Wallerman O, Wallén Arzt E, Berggren O, Elvers I, Eloranta ML, Rönnblom L, Lindblad Toh K, Wadelius C - Hum. Genet. (2016)

AS-SNPs associated with GWAS-SNPs. aTop AS-SNPs associated with GWAS-SNPs intersecting the cell-specific collections of AS-SNPs with the full GWAS catalog. Numbers are reported for common and rare AS-SNPs that are direct hits in the GWAS catalog or in LD with GWAS-SNPs. Bottom AS-SNPs associated with GWAS-SNPs intersecting the cell-specific collections of AS-SNPs with GWAS-SNPs associated with cell-specific traits. b Model representation of the networks of interactions observed between AS-SNPs and GWAS-SNPs. The dotted red box highlights the simplest scenario with one AS-SNP in LD with a GWAS-SNP. The tables report the numbers of instances observed in each cell line where one AS-SNP is in LD with several GWAS-SNPs or one GWAS-SNPs is in LD with different AS-SNPs, or where GWAS-SNPs were also AS-SNPs. c Four AS-SNPs, located in three different regulatory elements, interact with several GWAS-SNPs associated with autoimmune diseases
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4835527&req=5

Fig3: AS-SNPs associated with GWAS-SNPs. aTop AS-SNPs associated with GWAS-SNPs intersecting the cell-specific collections of AS-SNPs with the full GWAS catalog. Numbers are reported for common and rare AS-SNPs that are direct hits in the GWAS catalog or in LD with GWAS-SNPs. Bottom AS-SNPs associated with GWAS-SNPs intersecting the cell-specific collections of AS-SNPs with GWAS-SNPs associated with cell-specific traits. b Model representation of the networks of interactions observed between AS-SNPs and GWAS-SNPs. The dotted red box highlights the simplest scenario with one AS-SNP in LD with a GWAS-SNP. The tables report the numbers of instances observed in each cell line where one AS-SNP is in LD with several GWAS-SNPs or one GWAS-SNPs is in LD with different AS-SNPs, or where GWAS-SNPs were also AS-SNPs. c Four AS-SNPs, located in three different regulatory elements, interact with several GWAS-SNPs associated with autoimmune diseases
Mentions: The GWAS catalog has entries for a wide spectrum of traits that are expressed in different cell types, all of which were not studied in this experiment. Molecular events that could be explained by the cells investigated are for example immune-mediated diseases for GM12878 and K562, and neurological diseases for SK-N-SH (Table S3 in Supplementary material 1). Consequently, we searched the catalog for these cell-specific traits. H1-hESC was not investigated. The SNP with the strongest association (GWAS top hit) was collected from the catalog and SNPs in high LD (r2 > 0.8) were identified and intersected with the collection of AS-SNPs. For 36 traits, we found 141 AS-SNPs. We investigated the number of candidate regulatory SNPs found when searching the GWAS catalog using matched random sets of non-AS-SNPs. We found significantly more AS-SNPs than expected by chance with a clear enrichment (5- to 13-fold) in the overlap of AS-SNPs as compared to random sets of non-AS-SNPs (Fig S3 in Supplementary material 1). Out of the 141 unique AS-SNPs that were candidates to explain GWAS signals, only 15 were the particular SNP reported in the GWAS catalog and the other 126 were in the high-LD interval and 10 of these were rare AS-SNPs (Table 2; Fig. 3a, b and Tables S8, S11, S13 in Supplementary material 3). We grouped the SNPs in 1 Mb loci, and in immune cells, GM12878 and K562, we found candidate functional SNPs at 71 loci and in SK-N-SH at 14 loci, giving a total of functional candidates at 85 loci. We compared our candidates with the total number of mapped regions for different diseases and detected tentative functional variants for 11 % of loci mapped for psoriasis, 9 % for SLE and self-reported allergy, 8 % for inflammatory bowel disease/Crohn’s disease/ulcerative colitis and 6 % for type 1 diabetes, suggesting new functional regulatory elements for many common diseases.Table 2

Bottom Line: We found 9962 candidate regulatory SNPs, of which 16 % were rare and showed evidence of larger functional effect than common ones.Functionally rare variants may explain divergent GWAS results between populations and are candidates for a partial explanation of the missing heritability.Furthermore, by examining GWAS loci we found >400 allele-specific candidate SNPs, 141 of which were highly relevant in our cell types.

View Article: PubMed Central - PubMed

Affiliation: Science for Life Laboratory, Department of Immunology, Genetics and Pathology, Uppsala University, Uppsala, Sweden.

ABSTRACT
Genome-wide association studies (GWAS) have identified a large number of disease-associated SNPs, but in few cases the functional variant and the gene it controls have been identified. To systematically identify candidate regulatory variants, we sequenced ENCODE cell lines and used public ChIP-seq data to look for transcription factors binding preferentially to one allele. We found 9962 candidate regulatory SNPs, of which 16 % were rare and showed evidence of larger functional effect than common ones. Functionally rare variants may explain divergent GWAS results between populations and are candidates for a partial explanation of the missing heritability. The majority of allele-specific variants (96 %) were specific to a cell type. Furthermore, by examining GWAS loci we found >400 allele-specific candidate SNPs, 141 of which were highly relevant in our cell types. Functionally validated SNPs support identification of an SNP in SYNGR1 which may expose to the risk of rheumatoid arthritis and primary biliary cirrhosis, as well as an SNP in the last intron of COG6 exposing to the risk of psoriasis. We propose that by repeating the ChIP-seq experiments of 20 selected transcription factors in three to ten people, the most common polymorphisms can be interrogated for allele-specific binding. Our strategy may help to remove the current bottleneck in functional annotation of the genome.

No MeSH data available.


Related in: MedlinePlus