protein_quant: combine different emeta_cols values mapping to a protein

[evanglass]

Putting this here to be a bit more visible, but there are still a couple things that need to be addressed:

• Wording of the warning
• What to do with isoform e_meta version
• Ensure tests pass

Example test cases:

# This code should succeed without any warnings, with column B remaining in e_meta
mypepData <- group_designation(omicsData = pep_object, main_effects = c("Phenotype"))
mypepData = edata_transform(omicsData = mypepData, "log2")

imdanova_Filt <- imdanova_filter(omicsData = mypepData)
mypepData <- applyFilt(filter_object = imdanova_Filt, omicsData = mypepData, min_nonmiss_anova = 2)

imd_anova_res <- imd_anova(omicsData = mypepData, test_method = 'comb', pval_adjust_a_multcomp = 'bon', pval_adjust_g_multcomp = 'bon')

mypepData$e_meta$B <- sapply(mypepData$e_meta$RazorProtein, nchar)

results <- protein_quant(pepData = mypepData, method = 'rollup', combine_fn = 'median', emeta_cols = "B")
print(names(results$e_meta))

# This code should give a warning and remove the A column from e_meta
mypepData <- group_designation(omicsData = pep_object, main_effects = c("Phenotype"))
mypepData = edata_transform(omicsData = mypepData, "log2")

imdanova_Filt <- imdanova_filter(omicsData = mypepData)
mypepData <- applyFilt(filter_object = imdanova_Filt, omicsData = mypepData, min_nonmiss_anova = 2)

imd_anova_res <- imd_anova(omicsData = mypepData, test_method = 'comb', pval_adjust_a_multcomp = 'bon', pval_adjust_g_multcomp = 'bon')

mypepData$e_meta$A <- round(runif(nrow(mypepData$e_meta)), 1)*10

results <- protein_quant(pepData = mypepData, method = 'rollup', combine_fn = 'median', emeta_cols = "A")
print(names(results$e_meta))

# This code should give a warning about and remove column A, but still keep the B column in the e_meta
mypepData <- group_designation(omicsData = pep_object, main_effects = c("Phenotype"))
mypepData = edata_transform(omicsData = mypepData, "log2")

imdanova_Filt <- imdanova_filter(omicsData = mypepData)
mypepData <- applyFilt(filter_object = imdanova_Filt, omicsData = mypepData, min_nonmiss_anova = 2)

imd_anova_res <- imd_anova(omicsData = mypepData, test_method = 'comb', pval_adjust_a_multcomp = 'bon', pval_adjust_g_multcomp = 'bon')

mypepData$e_meta$A<- round(runif(nrow(mypepData$e_meta)), 1)*10
mypepData$e_meta$B<- sapply(mypepData$e_meta$RazorProtein, nchar)

results <- protein_quant(pepData = mypepData, method = 'rollup', combine_fn = 'median', emeta_cols = c("A", "B"))
print(names(results$e_meta))

# This code should also give a warning about column A and keep column B, but should use isoforms.
mypepData <- group_designation(omicsData = pep_object, main_effects = c("Phenotype"))
mypepData = edata_transform(omicsData = mypepData, "log2")

imdanova_Filt <- imdanova_filter(omicsData = mypepData)
mypepData <- applyFilt(filter_object = imdanova_Filt, omicsData = mypepData, min_nonmiss_anova = 2)

imd_anova_res <- imd_anova(omicsData = mypepData, test_method = 'comb', pval_adjust_a_multcomp = 'bon', pval_adjust_g_multcomp = 'bon')

isoformRes = bpquant(statRes = imd_anova_res, pepData = mypepData)

mypepData$e_meta$A<- round(runif(nrow(mypepData$e_meta)), 1)*10
mypepData$e_meta$B<- sapply(mypepData$e_meta$RazorProtein, nchar)

protein_quant(pepData = mypepData, method = 'rollup', combine_fn = 'median', isoformRes = isoformRes, emeta_cols = c("A", "B"))

[clabornd]

Ran through all the examples and looks good to me. For the tests, basically the construction of the 'standards' needs to be updated to have the emeta_cols included. The original tests just seem to ignore this.

e.g. the construction of stan_mea

pmartR/tests/testthat/test_quant_protein.R

Line 117 in ae81337

stan_mea <- as.proData(

needs to include a column 'Peptide_Sequence'

[clabornd]

@evanglass I wanted to get these changes in before I do a release, added one small change that I think resolves the test failures. I believe it was that rows for the extra emeta_cols were being collapsed before the counting of peps_per_pro/n_peps_used could be done.

Nothing to do on your part I think unless you see a problem with my commit.