update ld_matrix_local to create variant ID and standardise

[explodecomputer]

Update https://github.com/MRCIEU/ieugwasr/blob/master/R/ld_matrix.R#L92

• extract by chr:pos
• Rows and columns are variant ID
• r values in matrix are flipped according to variant ID.

Example:

  harmonise_ld <- function(d, rho)  {
    toflip <- d$effect_allele.exposure < d$other_allele.exposure
    d$beta.exposure[toflip] <- d$beta.exposure[toflip] * -1
    d$beta.outcome[toflip] <- d$beta.outcome[toflip] * -1
    temp <- d$effect_allele.exposure[toflip]
    d$effect_allele.exposure[toflip] <- d$other_allele.exposure[toflip]
    d$other_allele.exposure[toflip] <- temp
    rhoid <- do.call(rbind, strsplit(rownames(rho), split = "_")) |> as.data.frame(stringsAsFactors=FALSE)
    toflip2 <- rhoid$V2 < rhoid$V3
    temp <- rhoid$V2[toflip]
    rhoid$V2[toflip] <- rhoid$V3[toflip]
    rhoid$V3[toflip] <- temp
    rho[toflip, ] <- rho[toflip, ] * -1
    rho[, toflip] <- rho[, toflip] * -1
    return(list(d=d, rho=rho))
  }

[JonnyBaseball]

Thanks! I tried to do the same but not as elegantly! To be clear, this not only aligns alleles with the reference 1000G but also returns the ld matrix in same order as input data.