mrc-ide / mipanalyzer

Filtering and analysis of MIP data

MIPanalyzer

MIPanalyzer is a tool for analyzing MIP (molecular inversion probe) data. It performs the following functions:

• Reading in raw data, usually in the form of a vcf file produced by MIPwrangler
• Filtering samples and loci based on coverage, missingness, and other metrics
• Performing simple population genetic analyses, such as PCA and basic pairwise distance metrics

All documentation, including installation instructions and tutorials, are on the MIPanalyzer website.

Version History

The current release version of the software is v1.1.0, released 12 Dec 2023.

Contributors

• Bob Verity
• OJ Watson
• Nick Brazeau

mipanalyzer's Issues

Upset Plot

Goal to implement UpSet plots (https://upset.app/)

Code from Nat Comms here: https://github.com/bobverity/antimalarial_resistance_DRC/tree/master/figure6_UpSet

REHH Extension

Goal to implement REHH type plots within MIPanalyzer.

Code from Nat Comms paper here: https://github.com/nickbrazeau/mip_bigbarcode_ehh

Code was analysis-based, not package-based and will need some significant lifting for general use.

Extension for MAPs

Right now we have GC distance calculations but otherwise few spatial functions/capabilities. Worth extending?

Do we want spatial IBD networks?

sf and shape files are not small tasks... but is useful for comm?

array limited to 3 ALT

Hey Bob,

For the array under mipanalyzer_data your dimensions are dim(3, n, L) where 3 is ALT alleles, n is samples, L is loci. This absolutely makes sense for SNP data (where ALT is bounded by the 4 nucleotide bases) but will fail for INDEL data where the theoretical upper limit of ALT is n-1 (i.e. ever sample has its own unique INDEL). Granted, these latter points get a bit muddled, as we can see in example 5.1.2 here.
Regardless, I thought it was important to perserve the fact that ALT length changes depending on the loci in our data-structure. If you disagree and want to use an array of max(ALT) x n x L we can do that too?
Also -- sorry if I missing something -- but it was unclear to me why coverage was an array? I just returned a simple matrix ?

This is my vcf2mip_data function PR/proposa:

vcf2mipanalyzer_data <- function(file = NULL, vcfR = NULL, verbose = TRUE){
  #.....................
  # Read and check input
  #.....................
  if(is.null(file) & is.null(vcfR)){
    stop("You must specify an input: either a raw vcf file path or a vcfR object")
  } else if(!is.null(file) & !is.null(vcfR)){
    stop("You must specify one input: either a raw vcf file path or a vcfR object")

  } else if(!is.null(vcfR)){ # user specified a vcfR object
    if(!inherits(vcfR, "vcfR")){
      stop("vcfR object must be of class vcfR")
    }
    vcf <- vcfR

  } else if(!is.null(file)){ # user specified a file
    if(!file.exists(file)){
      stop("The vcf does not appear to exist at that file path")
    }
    vcf <- vcfR::read.vcfR(file=file, verbose=verbose) # read vcf
  }

  #..........................
  # find features and convert
  #..........................
  countmat <- vcfR::extract.gt(vcf, element="AD")
  counts <- sapply(1:nrow(countmat), function(x){ # R version of a for loop
    ret <- t(stringr::str_split(countmat[x,], ",", simplify = T))
    ret <- apply(ret, 2, as.numeric) # need to coerce to numeric here b/c of way vcfR extracts AD (can't set as.numeric T and retain all info) and to perserve df structure
    return(ret)
  }) # note, doing this way instead of an array to perserve the fact that the ALT count varies by loci as you could see by the command below
     # d <- sapply(vcf@fix[,5], function(x){ return(length(unlist(stringr::str_split(x, ","))))}) # ALT count per loci

  coverage <- vcfR::extract.gt(vcf, element="DP", as.numeric = T)

  # create data
  ret <- list(coverage = coverage,
              counts = counts,
              samples = colnames(vcf@gt)[2:ncol(vcf@gt)],
              loci = vcf@fix # keep all the info here not just loci pos?
              )

  # return in mipanalyzer_data class
  class(ret) <- "mipanalyzer_data"
  return(ret)
}

MIPanalyzer installation not working

im unable to install the MIPanalyzer using
devtools::install_github("mrc-ide/MIPanalyzer", ref = "v1.1.0")