hhabra/metabcombiner

metabCombiner R Package: Paired Untargeted Metabolomics Feature Matching & Data Concatenation

metabcombiner's Issues

update method should be exported

Update method should be exported for the purposes of metabCombiner re-analyses

Strip spaces at the beginning or end of strings

Strings are treated as different when the beginning or end of the string has spaces

difference in labelRows() output between v 1.2.2 and 1.5.2+

Hi,

I've been doing quite a bit of modeling with metabCombiner and noticed that with my recent biocmanager:: update, my old code stopped working.

Before, labelRows(method = "mzrt") worked on my combinedTable() dataframe

Now, i get Error in labelRows(ct_tp, method = "mzrt", delta = c(de1, de2, de3, de4), :
must call calcScores before using this function

The former behavior seems more intuitive to me since I am not using method="scores".

I'm not sure if I should treat my old modeling results as bad or try to replicate them with certain calcScores() parameter values though I don't know what these would be.

Do you have any insight or direction for me on this question?

outlier handling in plot_fit

New outlier argument should be set to either "none", "remove", "highlight"

@ symbol still in formCombinedTable

Get rid of those, need to use accessor method

Update anchors table in updateTables

Option to edit and insert anchors table in updateTables function

how to make the example data work

Hi,
I found your package recently and think it looks interesting for our research, while when I tried to run you script, I met a couple of issues. Firstly , while is the file "path_to_data_file.txt"? Unfortunately I did not find it even I installed you packages. Secondly, do you have a full scripts than can fully work for your example data, I found there are many symbols such as "..." in your script that not recognized by R and I could not run these codes for the example data. I would like to see what is the output of your package and whether we can apply to our data and how should we prepare our data format.

Thank you very much in advance.
Best
Tingting

calcScores group number bug

When performing a re-analysis of combined data, group numbers from 1 to nGroups will cause an error due to missing groups. Solution: use unique().

slow table collapsing in write2file() function

Current method for table collapsing in write2file function too slow
lines = apply(cTable, 1, function(row) base::paste(row, collapse = sep))

Faster methods necessary. One possibility:
strsplit(readr::format_delim(cTable, delim = sep), split = "\n")[[1]])

Also, addition of an index column may be useful and necessary.

later dplyr version needed

dplyr::all_of called in metabData(), but this functionality is only available in later versions of dplyr (starting from version 1.0). Update DESCRIPTION to set minimum dplyr to 1.0.

No getExtra method for metabCombiner objects

A method should be added

ability to access filtered metabData features

Features filtered out by metabData need to be accessible to the user. New slot for "filtered" features should be added and method "filtered" for retrieving the features.

detect regular expressions bug

Program may not recognize literal bracket characters ("(", "[", "{") in column names during fields detection. Solution is to substitute a period character for any of the above characters.

use match() instead of grep() in formCombinedTable()

In the file form.R, the function formCombinedTable() is causing an error for us because it uses grep() to get the indices of matching IDs. The grep is failing because the string for regex pattern gets too big since we have thousands of IDs.

A fix would be to use match() instead, which also avoid another potential source of error -see below. Here are my suggested changes, which seem to work, in the file form.R:

formCombinedTable <- function(object, xset, yset, xreps, yreps){
    xComb <- dplyr::slice(xset, rep(seq(1,n()), times = xreps))
    yComb <- yset[yreps,]
    samps_extras <- unlist(lapply(datasets(object), function(d)
        c(getSamples(object, d), getExtra(object,d))))
    # samps_extras <- rmbrackets(paste(samps_extras, collapse = "|"))            # removed
    # x_sa_ext <- setdiff(grep(samps_extras, rmbrackets(names(xComb))),seq(1,5)) # removed
    x_sa_ext <- na.omit(match(samps_extras, names(xComb)))                       # added
    x_sa_ext <- x_sa_ext[x_sa_ext > 5]                                           # added
    # y_sa_ext <- setdiff(grep(samps_extras, rmbrackets(names(yComb))),seq(1,5)) # removed
    y_sa_ext <- na.omit(match(samps_extras, names(yComb)))                       # added
    y_sa_ext <- y_sa_ext[y_sa_ext > 5]                                           # added
    cTable <- data.frame(idx = xComb[["id"]], idy = yComb[["id"]],
                    mzx = round(xComb[["mz"]],5), mzy = round(yComb[["mz"]],5),
                    rtx = round(xComb[["rt"]],4), rty = round(yComb[["rt"]],4),
                    rtProj = numeric(nrow(xComb)),
                    Qx = round(xComb[["Q"]],4), Qy = round(yComb[["Q"]],4),
                    group = xComb[["group"]],
                    score = rep(1, nrow(xComb)),
                    rankX = as.integer(1), rankY = as.integer(1),
                    adductx = xComb[["adduct"]], adducty = yComb[["adduct"]],
                    xComb[,sort(x_sa_ext)], yComb[,sort(y_sa_ext)],
                    stringsAsFactors = FALSE, check.names = FALSE)
    consec <- lapply(rle(cTable[["group"]])[["lengths"]], function(l) seq(1,l))
    consec <- unlist(consec)
    rowID <- sprintf("%s.%s", cTable[["group"]], consec)
    cTable <- cbind.data.frame(rowID = rowID, cTable)

    return(cTable)
}

The grep() approach will produce unintended matches in cases of varying length IDs where some IDs are substrings of others, for example, if the search pattern has the ID 123 then it will match 1230 1123 and 1231 etc.

Caesar

filterRT - use indexing instead of dplyr::filter

dplyr::filter mixes up the row indices

Add a combinedTable Check for Missing Values in Group

Currently outputs an unhelpful error when object check is made

QCol bug in detectFields

In detectFields, there is a -Q for the Q column deletion rather -QCol.

Group 152 Isomeric pair

A group in the Jen dataset has a first isomer labeled, but the second is labeled "remove"