Tax_glom over aggregating beyond specified taxonomy

[maxnoblemm]

Hi all,

Has anyone encountered an issue where tax_glom() appears to aggregate beyond the specified taxonomic level?

Problem description
I'm working with relative abundance data constructed from MetaPhlAn output. When I agglomerate by a specific taxonomic level, the resulting column sums—which should be ~1 if properly normalized—are much higher. For example:

ps_phylum <- tax_glom(physeq_obj, taxrank = "Phylum")  # relative abundance data

This results in column sums of 7 at the Phylum level, 6 at Class, 5 at Order, etc.—suggesting values may be summed across lower levels rather than aggregated appropriately. However, inspecting each taxonomic level individually confirms that the original data sums to ~1 per sample, as expected for relative abundance.

I'm wondering if this behavior could be due to how I constructed the phyloseq object, especially since:

• I only have relative abundance data from MetaPhlAn (no raw counts)
• There's no tree
• The object includes just the otu_table, tax_table, and sample_data

# Formating relative abundance table

mpa_phyloSeq <- mpa_data / 100  # Convert from percentage to proportion
Step 2: Prepare sample metadata

metadata_metaG_phylo <- metadata_metagenomics %>% 
  dplyr::select(-Diagnosis, -CD_onset, -Country) %>%
  inner_join(Expsm15mo_CaseID, by = "CaseID") %>%
  column_to_rownames(var = "CaseTime")

correct_samples <- intersect(rownames(mpa_phyloSeq), rownames(metadata_metaG_phylo))
mpa_phyloSeq_filtered <- mpa_phyloSeq[correct_samples, , drop = FALSE]
metadata_metaG_phylo <- sample_data(metadata_metaG_phylo)

# Build OTU table

mpa_phyloSeq_filtered <- t(mpa_phyloSeq_filtered)  # taxa as rows
OTU <- otu_table(mpa_phyloSeq_filtered, taxa_are_rows = TRUE)
Step 4: Construct tax_table from MetaPhlAn strings

otu_df <- as.data.frame(otu_table(physeq_MetaG))
tax_strings <- rownames(otu_df)
tax_split <- str_split(tax_strings, "\\|")
max_levels <- max(sapply(tax_split, length))

tax_split_padded <- lapply(tax_split, function(x) {
  length(x) <- max_levels
  return(x)
})

tax_matrix <- do.call(rbind, tax_split_padded)
colnames(tax_matrix) <- c("Kingdom", "Phylum", "Class", "Order", "Family", "Genus", "Species", "Strain")[1:max_levels]
tax_matrix <- apply(tax_matrix, 2, str_trim)

tax_tab <- tax_table(tax_matrix)
rownames(tax_tab) <- rownames(otu_df)
Step 5: Assemble final phyloseq object

physeq_obj <- phyloseq(OTU, metadata_metaG_phylo, tax_tab)

Request for help
When I then take physeq_obj and try to tax_glom as follows:

ps_phylum<- tax_glom(physeq_obj, taxrank = "Phylum")
head(ps_phylum@otu_table)

I get:

OTU Table:          [6 taxa and 533 samples]
                     taxa are rows
                                     CASE-1_12 CASE-1_15 CASE-1_18 CASE-1_21 
k__Bacteria|p__Actinobacteria        5.3338845 2.5500895 2.1597526 1.8547148 
etc...

Which shouldn't be possible as relative abundances are confined to sum to 1. I think somehow, my tax_tab isn't pulling just phyla data, and is including the lower level taxa data?

Any insights or suggestions would be hugely appreciated! Thank you!