How to construct your own exposure and outcome datasets?

[mocksu]

Suppose you have your own exposure GWAS and your own outcome GWAS, how can you do a 2 sample MR?

[remlapmot]

You can do this.

You need to put your exposure and outcome data into data.frames with the relevant column names. See the example below for the names or there is a little example here.

library(TwoSampleMR)
#> TwoSampleMR version 0.5.6 
#> [>] New: Option to use non-European LD reference panels for clumping etc
#> [>] Some studies temporarily quarantined to verify effect allele
#> [>] See news(package='TwoSampleMR') and https://gwas.mrcieu.ac.uk for further details
ao <- available_outcomes()
#> API: public: http://gwas-api.mrcieu.ac.uk/
#> API: public: http://gwas-api.mrcieu.ac.uk/
bmi_exp_dat <- extract_instruments(outcomes='ieu-a-2')
chd_out_dat <- extract_outcome_data(snps = bmi_exp_dat$SNP, outcomes = 'ieu-a-7')
#> Extracting data for 79 SNP(s) from 1 GWAS(s)
#> Extracting data for 79 SNP(s) from 1 GWAS(s)

class(bmi_exp_dat)
#> [1] "data.frame"
names(bmi_exp_dat)
#>  [1] "pval.exposure"          "samplesize.exposure"    "chr.exposure"          
#>  [4] "se.exposure"            "beta.exposure"          "pos.exposure"          
#>  [7] "id.exposure"            "SNP"                    "effect_allele.exposure"
#> [10] "other_allele.exposure"  "eaf.exposure"           "exposure"              
#> [13] "mr_keep.exposure"       "pval_origin.exposure"   "data_source.exposure"

class(chd_out_dat)
#> [1] "data.frame"
names(chd_out_dat)
#>  [1] "SNP"                   "chr"                   "pos"                  
#>  [4] "beta.outcome"          "se.outcome"            "samplesize.outcome"   
#>  [7] "pval.outcome"          "eaf.outcome"           "effect_allele.outcome"
#> [10] "other_allele.outcome"  "outcome"               "id.outcome"           
#> [13] "originalname.outcome"  "outcome.deprecated"    "mr_keep.outcome"      
#> [16] "data_source.outcome"

^{Created on 2021-07-26 by the reprex package (v2.0.0)}

You should then be able to run harmonise_data() on your data.frames as per our examples, e.g. here.