Introduction. inthinnerator is a command-line utility to help with the common task of "thinning" genetic variants, most commonly SNPs, from a dense (usually genome-wide) set of variants such as you would use in a GWAS study. Inthinnerator might be useful if:

• You want an approximately independent set of variants for use in other analyses (like principal components analysis).
• Or you want to pick out 'top' loci (e.g. lead GWAS SNPs) as best representatives of variation in each region.
• Or you want to simulate 'null' sets of variants to use, for example, in empirical enrichment analysis.

Features. Inthinnerator can be used to:

• Pick SNPs by rank (e.g. association P-value), randomly among SNPs, or uniformly in the genome.
• Thin variants based on physical distance, recombination distance, or a combination of both.
• Annotate variants with physical and recombination distances, and physical or recombination regions.
• Annotate variants with nearby genes.

Note: Inthinnerator does not currently thin variants based on estimates of LD, though we may implement that in future. For LD-based thinning, try other programs such as plink.

Change history.
Important! This page documents version 2 of inthinnerator which is currently experimental. This means we expect some features not to work, or not to work well, or to work wrongly, or to destroy your computer or sanity.

References. In different incarnations inthinnerator has been used in the guts of several papers. Here are a few:

1. Band et al, "Imputation-Based Meta-Analysis of Severe Malaria in Three African Populations", PLOS Genetics (2013)
2. Su et al, "Common variants at the MHC locus and at chromosome 16q24.1 predispose to Barrett's esophagus", Nature Genetics (2012)

Acknowledgements. The following people contributed to the design and implementation of qctool:

• Gavin Band

Contact. For more information or questions, please contact the oxstatgen mailing list at

                        oxstatgen (at) jiscmail.ac.uk
                    

The general process followed by inthinnerator is shown on the right. In brief, inthinnerator picks SNPs by alternating two processes:

1. Pick a SNP from the set of SNPs remaining, using a picking strategy.
2. Exclude SNPs from the remaining list using an thinning strategy (typically by excluding SNPs in a physical or recombination interval around the picked SNP.)
3. Either go back to step 1 and repeat, or stop and write the results.

Picking. Inthinnerator implements the following strategies for picking variants in step 1:

• Pick a SNP at random from the SNPs remaining (-strategy random). This is the default behaviour.
• Pick a SNP at uniformly at random from regions of the genome covered by SNPs in the input file (-strategy random_by_position). The -bin-size option controls how genomic coverage is computed.
• Pick the first SNP remaining (-strategy first).
• Pick SNPs according to a specified rank (-rank).

Thinning. Inthinnerator implements the following strategies for thinning variants in step 2:

• Thin by physical distance, e.g. -min-distance 100kb, -min-distance 20bp, -min-distance 2.5Mb.
• Thin by recombination distance, e.g. -min-distance 0.125cM.
• Thin by recombination distance with a physical margin, e.g. -min-distance 0.125cM+25kb.

Stopping. Inthinnerator can choose to stop picking in a few ways:

• Continue until there are no SNPs left to pick (this is the default behaviour).
• Continue until a specified number of SNPs have been picked (-max-picks).
• Continue until one SNP has been picked for each tag specified to -match-tag.

Repeating. Specifying -N x, where x is a number > 1, will cause inthinnerator to repeat the whole thinning process x times. Output files will be numbered from 0 to x-1.

By default inthinnerator will write all variants included and excluded from the thinned list to the output file. The options -suppress-excluded and -suppress-included can be used to adjust this behaviour.

Example output. Basic tab-delimited inthinnerator output looks like this:

# Analysis: "inthinnerator analysis"
#  started: 2014-09-02 08:54:33
# 
# Analysis properties:
#   -g ../imputed_chr22.index (user-supplied)
#   -map ../genetic_map_chr#_combined_b37.txt (user-supplied)
#   -o example (user-supplied)
# 
alternate_ids   rsid    chromosome      position        alleleA alleleB iteration       pick_index      result  cM_from_start_of_chromosome     region_lower_bp region_upper_bp region_lower_cM region_upper_cM
?       kgp14987749     22      16152031        A       C       0       2676    picked  0.462220808233324       16150629        16153432        0.452218357350001       0.472216124700895
---     22-16156144     22      16156144        GC      G       0       1855    picked  0.491564660218421       16154742        16157545        0.481562209335098       0.501559976685992
---     22-16158548     22      16158548        ACT     A       0       1325    picked  0.508715795684546       16157146        16159949        0.498713344801223       0.518711112152117
---     22-16160493     22      16160493        A       AC      0       4950    picked  0.522592234320824       16159091        16161894        0.512589783437502       0.532587550788396
---     22-16163055     22      16163055        TTATC   T       0       3397    picked  0.540870607475655       16161653        16164456        0.530868156592332       0.550865923943226
---     22-16164909     22      16164909        CCT     C       0       4524    picked  0.554097814278565       16163507        16166310        0.544095363395242       0.564093130746136