Question

BiocCheck object initialization - NOTE

0

Entering edit mode

ullrich ▴ 20

@ullrich-23697

Last seen 2.5 years ago

Germany

Hi,

the BiocChek gives a NOTE about NOTE: Consider clarifying how 9 object(s) are initialized. they are part of a data set loaded with data(), or perhaps part of an object referenced in with() or within(). function (object) codonmat2xy (Codon) codonmat2xy (syn) codonmat2xy (nonsyn) codonmat2xy (SynSum) codonmat2xy (NonSynSum) codonmat2xy (IndelSum) codonmat2xy (SynMean) codonmat2xy (NonSynMean) codonmat2xy (IndelMean)

However, the objects are initialized within a parallel foreach loop and than processed with dplyr.

Is there a way how to handle this kind of object initialization without getting a NOTE from BiocCheck?

Thank you in anticipation

Best regards

Kristian Ullrich

codonmat2xy <- function(codonmat, threads = 1){
  doMC::registerDoMC(threads)
  i <- NULL
  j <- NULL
  k <- NULL
  OUT <- foreach(i = seq(from = 1, to = nrow(codonmat)), .combine=rbind) %dopar% {
    foreach(j = seq(from = 1, to = ncol(codonmat) - 1), .combine=rbind) %do% {
      foreach(k = seq(from = j + 1, to = ncol(codonmat)), .combine=rbind) %do% {
        c(setNames(i, "Codon"),
          setNames(j, "Comp1"),
          setNames(k, "Comp2"),
          setNames(compareCodons(codonmat[i, j], codonmat[i, k]), c("syn", "nonsyn", "indel")))
      }
    }
  }
  OUT <- as.data.frame(OUT)
  OUT.NAs <- OUT %>% dplyr::group_by(Codon) %>% dplyr::filter(!is.na(syn)) %>%
                     dplyr::count(Codon)
  OUT.SynSum <- OUT %>% dplyr::group_by(Codon) %>%
                        dplyr::summarise(SynSum = sum(syn, na.rm = TRUE))
  OUT.NonSynSum <- OUT %>% dplyr::group_by(Codon) %>%
                           dplyr::summarise(NonSynSum = sum(nonsyn, na.rm = TRUE))
  OUT.IndelSum <- OUT %>% dplyr::group_by(Codon) %>%
                          dplyr::summarise(IndelSum = sum(indel, na.rm = TRUE))
  OUT.join <- dplyr::left_join(OUT.NAs, OUT.SynSum) %>%
             dplyr::left_join(OUT.NonSynSum) %>% dplyr::left_join(OUT.IndelSum)
  OUT.xy <- OUT.join %>% dplyr::mutate(SynMean = SynSum/n,
                                       NonSynMean = NonSynSum/n,
                                       IndelMean = IndelSum/n)
  OUT.xy <- OUT.xy %>% dplyr::mutate(CumSumSynMean = cumsum(SynMean),
                                       CumSumNonSynMean = cumsum(NonSynMean),
                                       CumSumIndelMean = cumsum(IndelMean))
  return(OUT.xy)
}

BiocC BiocCheck • 1.1k views

ADD COMMENT • link 3.5 years ago ullrich ▴ 20

score 0 · Answer 1 · 2021-06-01

0

Entering edit mode

Marcel Ramos 700

@marcel-ramos-7325

Last seen 2 days ago

United States

Hi Ullrich,

This is often the case when using NSE in dplyr. You can use utils::globalVariables at the top of your R file to declare these and avoid the NOTE.

Best regards,

Marcel

ADD COMMENT • link 3.5 years ago Marcel Ramos 700

0

Entering edit mode

Thank you Marcel,

I could solve it by adding #' @importFrom rlang .data

now the code looks like this:

codonmat2xy <- function(codonmat, threads = 1){
    cl <- parallel::makeForkCluster(threads)
    doParallel::registerDoParallel(cl)
    i <- NULL
    j <- NULL
    k <- NULL
    OUT <- foreach::foreach(i = seq(from = 1, to = nrow(codonmat)),
            .combine=rbind, .packages = c('foreach')) %dopar% {
        foreach::foreach(j = seq(from = 1, to = ncol(codonmat) - 1),
            .combine=rbind) %do% {
            foreach::foreach(k = seq(from = j + 1, to = ncol(codonmat)),
                .combine=rbind) %do% {
                c(setNames(i, "Codon"),
                    setNames(j, "Comp1"),
                    setNames(k, "Comp2"),
                    setNames(distSTRING::compareCodons(codonmat[i, j],
                        codonmat[i, k]), c("syn", "nonsyn", "indel")))
            }
        }
    }
    parallel::stopCluster(cl)
    OUT <- as.data.frame(OUT)
    OUT.NAs <- OUT %>% dplyr::group_by(.data$Codon) %>%
        dplyr::filter(!is.na(.data$syn)) %>%
        dplyr::count(.data$Codon)
    OUT.SynSum <- OUT %>% dplyr::group_by(.data$Codon) %>%
        dplyr::summarise(SynSum = sum(.data$syn, na.rm = TRUE))
    OUT.NonSynSum <- OUT %>% dplyr::group_by(.data$Codon) %>%
        dplyr::summarise(NonSynSum = sum(.data$nonsyn, na.rm = TRUE))
    OUT.IndelSum <- OUT %>% dplyr::group_by(.data$Codon) %>%
        dplyr::summarise(IndelSum = sum(.data$indel, na.rm = TRUE))
    OUT.join <- dplyr::left_join(OUT.NAs, OUT.SynSum) %>%
        dplyr::left_join(OUT.NonSynSum) %>% dplyr::left_join(OUT.IndelSum)
    OUT.xy <- OUT.join %>% dplyr::mutate(SynMean = .data$SynSum/.data$n,
        NonSynMean = .data$NonSynSum/.data$n,
        IndelMean = .data$IndelSum/.data$n)
    OUT.xy <- OUT.xy %>%
        tibble::add_column(CumSumSynMean = cumsum(OUT.xy$SynMean),
        CumSumNonSynMean = cumsum(OUT.xy$NonSynMean),
        CumSumIndelMean = cumsum(OUT.xy$IndelMean))
    return(OUT.xy)
}

ADD REPLY • link 3.5 years ago ullrich ▴ 20