R 在多个 data.frames 之间循环并为它们赋值

Question

I'm using R to perform some alterations in cnvkit output (for my purposes). The thing is: doing the job sample by sample, the script works like a charm but, when I put it into a for loop, it breaks!

Tried a lot of answers posted on Stack Overflow but none of then helped me.

# Clear workspace
rm(list=(ls()))

ref <- read.csv("/path/to/reference.cnn", header=T, sep="\t")
path <- "/path/to/call_files/"
files = list.files(path = path, pattern = "*.final.call.cnr", full.names=FALSE)
for(file in files) {
    perpos <- which(strsplit(file, "")[[1]]==".")
    assign(
    gsub(" ","",substr(file, 1, perpos-1)), 
    read.csv(paste(path,file,sep=""), header=T, sep="\t"))

}


mod_CNV = function(x) {

    # Merge both files by "start" position
    merged <- merge(files[i], ref, by="start", suffixes=c(".files[i]", ".ref"))

    # Round "log2" column
    merged$log2.D00893 <- round(merged$log2.files[i], digits=1)

    # re-calculate "cn" based on log2 correction
    merged$cn <- round(2*(2^(merged$log2.files[i])))

    # Subset file with all "cn" values that are not 2
    alt.cn <- subset(merged, merged$cn !=2)

    # Create new data with columns of interest
    alt.cns <- as.data.frame(alt.cn[, c(1:8,13)])

    # Re-order columns for better view
    alt.cns <- alt.cns[c(2,1,3,4,6,5,8,7,9)]

    # Calculate ratio between coverages
    alt.cns$depth.ratio <- round(alt.cns$depth.files[i] / alt.cns$depth.ref, digits=2)
    alt.cns$depth.ratio.1 <- round(alt.cns$depth.files[i] / alt.cns$depth.ref, digits=2)

    ## Function to call for DUP or DEL.  
    alt.cns$SV_type <- ifelse(alt.cns$cn < 2, "DEL", "DUP")

    # Convert "alt.cns" to .bed file
    full <- alt.cns[c(1,2,3,12,5,4,6,7,8,9,10)]
    names(full)[1] <- "#Chrom"
    names(full)[2] <- "Start"
    names(full)[3] <- "End"
    names(full)[4] <- "SV_type"
    names(full)[6] <- "gene"
    names(full)[7] <- "log2"

    # Save "alt.cns" as .bed file
    write.table(full, file="/path/to/output/files[i].bed", row.names=F, col.names=T, sep="\t")

    # Filter "alt.cns" file
    filtered <- subset(alt.cns, alt.cns$depth.ratio < 0.70 | alt.cns$depth.ratio > 1.40 & alt.cns$weight > 0.3)
    filtered <- filtered[c(1,2,3,12,5,4,6,7,8,9,10)]
    names(filtered)[1] <- "#Chrom"
    names(filtered)[2] <- "Start"
    names(filtered)[3] <- "End"
    names(filtered)[4] <- "SV_type"
    names(filtered)[6] <- "gene"
    names(filtered)[7] <- "log2"

    #Save file
    write.table(filtered, file="/path/to/output/files[i].bed", row.names=F, col.names=T, sep="\t")

}


for ( i in seq_along(files)) {
        mod_CNV(files[i])
    }

What I expect is that the loop reads file by file and assign each individual file name to variables files[i] and save as .pdf. But, I'm getting a error right on the beginning of the code:

"Error in fix.by(by.x, x) : 'by' must specify a uniquely valid column".

For some reason, the loop isn't recognizing my files[i] variable, which is causing this error. Can someone help me in this problem? To be clear, this error doesn't occur when running sample by sample, out of the loop.

Answer 1

Welcome to StackOverflow!

You've declared a function:

mod_CNV = function(x) {

    # Merge both files by "start" position
    merged <- merge(files[i], ref, by="start", suffixes=c(".files[i]", ".ref"))
    .
    .
    .
}

From what I can tell, there is no reason that this function should know what i is; this is probably why files[i] fails.

Here is where i is located

for ( i in seq_along(files)) {
    mod_CNV(files[i])
}

i is a variable that is local to the for loop. If you want it to be available inside mod_CNV, you'd need to pass it in as a parameter.

What you are passing in to mod_CNV is the filename. Inside of mod_CNV , this filename is referred to as x yet I don't see anywhere inside mod_CNV where you use x .

This is how you should declare your function and make use of the filename you are passing in:

mod_CNV = function(filename) {

    # Merge both files by "start" position
    merged <- merge(filename, ref, by="start", suffixes=c(filename, ".ref"))
    .
    .
    .
    # replace all other occurrences of `file[i]` with `filename`
}

And you can loop through the list of files and call mod_CNV like this, without using i :

for (file in files) {
    mod_CNV(file)
}

Also, I haven't used merge before and I don't know exactly what you are trying to do... but I find it odd to use an entire filename as a suffix. But it may be what you intended.

Anyway, this should be enough information for you to resolve your issue.

Answer 2

For those who falls on the same problem as I, there goes the right code:

path <- "/path/to/files/"
files = list.files(path = path, pattern = "*.file.ext", full.names=FALSE)
for(file in files) {
    perpos <- which(strsplit(file, "")[[1]]==".")
    assign(
    gsub(" ","",substr(file, 1, perpos-1)), 
    read.csv(paste(path,file,sep=""), header=T, sep="\t"))

}

s_ref <- read.csv("/read/ref/file", header=T, sep="\t")
s_ref["depth.ref.norm"] <- round(s_ref["depth"]/mean(s_ref[["depth"]]), digits=2)

mod_CNV = function(file) {
    file_df <- read.csv(file, header=T, sep="\t")

    # Normalize $depth by mean
    file_df[sprintf("depth.%s.norm", file)] <- round(file_df[["depth"]]/mean(file_df[["depth"]]), digits=2)

    # Merge both files by "start" position
    merged <- merge(file_df, s_ref, by="start", suffixes=c(sprintf(".%s", file), ".ref"), all=TRUE)

    # Round "log2" column
    log2_col_name = sprintf("log2.%s", file)
    merged[log2_col_name] <- round(merged[[log2_col_name]], digits=1)

    # re-calculate "cn" based on log2 correction
    merged["cn"] <- round(2*(2^(merged[[log2_col_name]])))

    # Subset file with all "cn" values that are not 2
    alt_cn <- subset(merged, merged[["cn"]] != 2)

    # Create new data with columns of interest
    alt_cns <- as.data.frame(alt_cn[, c(1:9,14,18)])

    # Re-order columns for better view
    alt_cns <- alt_cns[c(2,1,3,4,6,5,8,7,9,10,11)]

    # Calculate ratio between coverages
    alt_cns["depth.ratio.norm"] <- round(alt_cns[[sprintf("depth.%s.norm", file)]] / alt_cns[["depth.ref.norm"]], digits=2)

    alt_cns["depth.ratio"] <- round(alt_cns[[sprintf("depth.%s", file)]] / alt_cns[["depth.ref"]], digits=2)

    ## Function to call for DUP or DEL.  
    alt_cns["SV_type"] <- ifelse(alt_cns$cn < 2, "DEL", "AMP")

    # Convert "alt.cns" to .bed file
    full <- alt_cns[c(1,2,3,14,5,4,6,7,8,9,10,11,12,13)]
    names(full)[1] <- "#Chrom"
    names(full)[2] <- "Start"
    names(full)[3] <- "End"
    names(full)[4] <- "SV_type"
    names(full)[6] <- "gene"
    names(full)[7] <- "log2"

    full["weight"] <- round(full[["weight"]], digits = 2)
    full <- full[order(full$"#Chrom"),]

    # Save "full" as .bed file
    output_file = sprintf("/path/%s.bed", file)
    write.table(full, file=output_file, row.names=F, col.names=T, sep="\t", dec=",")

}
    print(files)
    for (file in files) {
        mod_CNV(file)
}