All examples work with test data from the ref and gwas folder, therefore use much smaller GWAS datasets and reference panels than in reality to limit computation time.
The minimal requirements are:
--gwas: path to the GWAS summary statistics text file, containing at least the following columns SNP-id, Z-statistic, reference allele and risk allele and at least one row,--ref: path to the reference panel,--out: path for output file.
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt
For more info regarding automatically column names recognition of the GWAS file, see section GWAS dataset in the manual.
ssimp --gwas gwas/small.random.txt.gz --ref ref/small.vcf.sample.vcf.gz --out output.txt
.gz works, .zip does not work.
If you have a reference panel REF/myrefpanel{CHR}.vcf.gz available, use it as follows
ssimp --gwas gwas/small.random.txt --ref REF/myrefpanel{CHR}.vcf.gz --out output.txt
For detailed instructions and explanations see detailed instructions in usage-text.
A quick solution is to run ssimp without reference panel, but with a shortcut indicating 1KG and a preferred population. This will create a folder called refpanel and download 1KG (all populations, not only the selected one).
ssimp --gwas gwas/small.random.txt --ref 1KG/EUR --out output.txt
Where EUR can be replaced with any (super) population of 1000 genomes reference panel.
Then, use it as follows:
ssimp --gwas gwas/small.random.txt ~/reference_panels/1000genomes/ALL.chr{CHRM}.phase3_shapeit2_mvncall_integrated_v5a.20130502.genotypes.vcf.gz output.txt --sample.names ~/reference_panels/1KG/integrated_call_samples_v3.20130502.ALL.panel/sample/super_pop=EUR
No special argument needed, but GWAS input file needs to contain effect size b along with the P-value p.
The header of the GWAS file should be: SNP a1 a2 b p
ssimp --gwas gwas/small.random.p.b.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt
No special argument needed, but GWAS input file needs to contain Z-statistics of the odds ratios.
Alternatively, if no Z-statistic is available, provide b=log(OR) and the P-value.
The header of the GWAS file should either be SNP a1 a2 Z or SNP a1 a2 b p.
First, impute Z-statistics as shown above, then
the Z-statistic into se(b) and b using this R-function.
Provide chromosome and position instead, but include a SNP column that is empty too.
ssimp --gwas gwas/small.random.chr.pos.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --impute.range 22:18000000-22:18100075
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --impute.range 22
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --impute.range 20-22
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --impute.snp gwas/listofimputesnps.txt
listofimputesnps.txt contains SNP id's separated by new lines (no header).
If it is only a few of SNPs it might be easier to use:
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --impute.snp <(echo rs2587101 rs2277831 | tr ' ' '\n')
A more complex example with SNPs from two different chromosomes:
ssimp gwas/GIANT_HEIGHT_Wood_et_al_2014_publicrelease_HapMapCeuFreq.txt.gz ref/sub1KG-tiny/chr{CHRM}.vcf.gz output.txt --sample.names ref/link.to.1kg.data/integrated_call_samples_v3.20130502.ALL.panel/sample/super_pop=EUR --impute.snps <(echo rs148911000 rs111659000 rs183059100 rs76979500 rs150095300 rs115012100 rs187649300 rs560286600 rs78808100 | tr ' ' '\n')
This was fixed in version 0.3.
ssimp --gwas gwas/small.random.x.txt --ref ref/sub1KG-tiny/chrX.vcf.gz --out output.txt
Impute single SNPs (rsid)
ssimp --gwas gwas/small.random.x.txt --ref ref/sub1KG-tiny/chrX.vcf.gz --out output.txt --impute.snp <(echo rs183055800 rs146115300 rs150092800 | tr ' ' '\n')
Impute range of SNPs (X:pos)
ssimp --gwas gwas/small.random.x.txt --ref ref/sub1KG-tiny/chrX.vcf.gz --out output.txt --impute.range 23:18000000-23:18100075
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --tag.snp gwas/listoftagsnps.txt
listoftagsnps.txt contains SNP id's separated by new lines (no header).
If it is only a few of SNPs it might be easier to use:
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --tag.snp <(echo rs2305001 rs10854521 | tr ' ' '\n')
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --sample.names ref/filename.samples.small.txt
filename.samples.small.txt contains sample id's separated by new lines (no header).
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --sample.names ref/filename.samples.pop.txt/sample/super_pop=EAS
Here, filename.samples.txt contains sample id's (sample) along with a second attribute (here super_pop) that has different values, among them is EUR, for which we separate.
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --lambda 0.01
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --impute.maf 0.05
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --tag.maf 0.05
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --window.width 500000 --flanking.width 100000
ssimp --gwas gwas/small.random.n.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --missingness dep
Your GWAS summary statistics needs to have a sample size column.
ssimp --gwas gwas/small.random.txt --ref ref/small.vcf.sample.vcf.gz --out output.txt --log my_ssimp_logfile
Check R-file doc/sanitycheck_reimputed.R
This is not formally implemented, but there is a workaround documented here.
Follow the instructions in doc/usage.