I'm trying to perform a one-sample MR analysis based on individual data provided by UK Biobank. I have performed an association test using PLINK2 which generated 2 summary statistics. For one phenotype there were over 700 SNPs that passed the significance threshold of 5e-8. I have pruned the SNPs based on linkage disequilibrium and kept only 1 significant SNP for each high-LD region, resulting in a set of 100 significant SNPs. How should I further select instrument variables?