The UKB cohort includes a number of related groups, including trios made up of two parents and an offspring. These relationships are not collected as part of the phenotypic data, but they have been inferred from the genotyping data. See Bycroft et al, which states
"By considering the relationship types and the age and sex of the individuals within each family group, we identified 1,066 sets of trios (two parents and an offspring), which comprise 1,029 unique sets of parents and 37 quartets (two parents and two children)."
I don't know whether any of these parents are heterozygous at the relevant locations.
Relationship file ukb_rel.dat , derived from the UK Biobank genotyping data, is available on the RAP in folder Bulk > Genotype Results > Genotype Calls . The same file is available to download from Showcase/AMS via the command gfetch rel, described in the notes section of Category:263 and in Resource:668 (Section 6). To download this data from AMS, your project must have a basket containing Field:22418 (Genotype Calls). The calls themselves do not need to be downloaded, it is just necessary to have this field in a basket to authorise the download of the relatedness file. The file is the output of the KING algorithm and gives HetHet, IBS0 and Kinship information which can then be used to identify the trios.
Please note that I am not an expert in genetics and the following suggestion is purely speculative. If anyone can make a better suggestion, please feel free to correct me.
If there are not enough heterozygous sets, you could make a rough estimate based on probability, and on the assumption that the crossover rate in chromosome 3 is on average the same within the target region as outside the target region:
Human chromosomes have between 1 and 4 crossovers ,
Chromosome 3 is ~ 198,000,000 base-pairs in length ,
Distance from the start of NLGN1 to the start of ADIPOQ is ~ 13,000,000 base-pairs ,
Thus the proportion of the chromosome where a crossover would be relevant is about 13/198 ~ 1/15 ~ 0.066 ,
If chr3 has only one crossover, the chance of it being in the target region is ~ 0.066 ,
If chr3 has four crossovers, the chance that exactly one crossover is in the target region is 4 x 1/15 x 14/15 x 14/15 x 14/15 ~ 0.219 ,
If chr3 has two or three crossovers, the chance will be something intermediate.
This estimate could be improved if you know whether there is a normal rate of hotspots within the target region, if you know the most likely number of crossovers for chromosome 3, or if you know the actual positions of the SNPs rather than using the start of the genes. You could also take into account the chance of there being 3 out of 4 crossovers within the target region, which I have ignored as it will be very small compared with the other errors in the estimate.
Comments
1 comment
The UKB cohort includes a number of related groups, including trios made up of two parents and an offspring. These relationships are not collected as part of the phenotypic data, but they have been inferred from the genotyping data. See Bycroft et al, which states
"By considering the relationship types and the age and sex of the individuals within each family group, we identified 1,066 sets of trios (two parents and an offspring), which comprise 1,029 unique sets of parents and 37 quartets (two parents and two children)."
I don't know whether any of these parents are heterozygous at the relevant locations.
Relationship file ukb_rel.dat , derived from the UK Biobank genotyping data, is available on the RAP in folder Bulk > Genotype Results > Genotype Calls . The same file is available to download from Showcase/AMS via the command gfetch rel, described in the notes section of Category:263 and in Resource:668 (Section 6). To download this data from AMS, your project must have a basket containing Field:22418 (Genotype Calls). The calls themselves do not need to be downloaded, it is just necessary to have this field in a basket to authorise the download of the relatedness file. The file is the output of the KING algorithm and gives HetHet, IBS0 and Kinship information which can then be used to identify the trios.
Please note that I am not an expert in genetics and the following suggestion is purely speculative. If anyone can make a better suggestion, please feel free to correct me.
If there are not enough heterozygous sets, you could make a rough estimate based on probability, and on the assumption that the crossover rate in chromosome 3 is on average the same within the target region as outside the target region:
Human chromosomes have between 1 and 4 crossovers ,
Chromosome 3 is ~ 198,000,000 base-pairs in length ,
Distance from the start of NLGN1 to the start of ADIPOQ is ~ 13,000,000 base-pairs ,
Thus the proportion of the chromosome where a crossover would be relevant is about 13/198 ~ 1/15 ~ 0.066 ,
If chr3 has only one crossover, the chance of it being in the target region is ~ 0.066 ,
If chr3 has four crossovers, the chance that exactly one crossover is in the target region is 4 x 1/15 x 14/15 x 14/15 x 14/15 ~ 0.219 ,
If chr3 has two or three crossovers, the chance will be something intermediate.
This estimate could be improved if you know whether there is a normal rate of hotspots within the target region, if you know the most likely number of crossovers for chromosome 3, or if you know the actual positions of the SNPs rather than using the start of the genes. You could also take into account the chance of there being 3 out of 4 crossovers within the target region, which I have ignored as it will be very small compared with the other errors in the estimate.
Please sign in to leave a comment.