Merge and split 2 phyloseq objects

[NinaKons]

Hi Joey,

I have just started looking at the phyloseq code and was wandering how can I merge and split 2 phyloseq objects based on the sampel_IDs in the metadata table?

Merge

• To make sure that the shared RSVs between the 2 datasets (16S DNA and cDNA) are removed given that the same RSVs are assigned the same taxonomy, but how can I check this?
• 2 datasets (16S DNA and cDNA) are comparable ie. have undergone the same phyloseq processing steps
• Same colours for the tax_rank, groups etc are used in the plots

Split or subset this common phyloseq object

• Analyze and visualize the 2 datasets separately in terms of all their samples and a subset of shared samples (based on 2 separate metadata tables)

Merge

Load 3 input files generated via DADA2 pipeline

seqtab.nochim <- readRDS("/home/ubuntu/work/cDNA/trimmed/seqtab.rds") # Chimer-removed RSV table tax <- readRDS("/home/ubuntu/work/cDNA/trimmed/tax.rds") # Taxonomy table mt <- read.csv(file="metadata.csv", header=TRUE) # Metadata

Phyloseq object

ps_DNA <- phyloseq(otu_table(seqtab.nochim_sp, taxa_are_rows = FALSE), tax_table(tax), sample_data(mt))

Remove unclassified taxa and reduce sparsity

physeq_DNA <- subset_taxa(ps_DNA, Kingdom != "k__unclassified") physeq_DNA <- prune_taxa(taxa_sums(ps_DNA) > 0 , ps_DNA)

Merge

Can apply the same code to cDNA dataset and try to merge the 2 phyloseq objects:

ps_merge <- merge_phyloseq(physeq_DNA, physeq_cDNA)

But, have not tested whether merge_phyloseq can merge 2 phyloseq objects or found this case in the tutorial (https://joey711.github.io/phyloseq/merge.html#merge_phyloseq).

• Can you please, confirm that merge_phyloseq indeed can merge 2 phyloseq objects?
• As you see my phyloseq object(s) were created without a tree, do you think it is going to cause problems during object merging, plotting bars and ordinations, and ADONIS analyses?

Subset

Subset DNA and cDNA with the same samples

sub_ps_same_samples = subset_samples(ps_merged, mt_DNA$Short_name = mt_cDNA$Short_name)

Split the subsetted phyloseq object

sub_ps_DNA = subset_samples(sub_ps_same_samples, Short_name = mt_DNA$Short_name) sub_ps_cDNA = subset_samples(sub_ps_same_samples, Short_name = mt_cDNA$Short_name)

Many thanks in advance!

[joey711]

@benjjneb do you have a suggestion here?

[joey711]

Yes, this is the intended use of merge_phyloseq. The sample names from the two phyloseq objects should not overlap. The taxa_names for the same RSV must be the same. Your sample_data should have a variable that indicates which assay type generated each sample (for your sanity).

The fact of not including a tree (yet) is fine, as there is no good way to merge two trees that I'm aware. If someone shows me one I'd love to support it, though. Once you have the final set of RSV sequences, you can then compute a tree and add to the combined phyloseq object.