Clustering dada2-obtained ASVs into OTUs with DECIPHER

[ConstanceBtd]

Hello,

I have a phyloseq object that I obtained with the dada2 pipeline. I would like to cluster those ASVs into OTUs. For that, I have the following code (just the start of it):

clusters<-DistanceMatrix(refseq(physeq_asv), includeTerminalGaps = T, processors=1)
clusters<-TreeLine(clusters,method="single", cutoff=0.03, processors=NULL)

with physeq_asv being my phyloseq object obtained with dada2.
For the second line I have the following error:

Error in DECIPHER::TreeLine(clusters_matrix, method = "single", cutoff = 0.03,  :
  myDistMatrix must be a matrix for method 'single'.

Does anyone know where it could come from ?

Thanks !

[benjjneb]

I'm not sure, this is an error coming from the DECIPHER package, but the developer of that package is @digitalwright and they might be able to shed more light.

One note is that the code block you posted and the code causing the error don't exactly match up, one is using the variable name clusters while the other is using the variable name clusters_matrix.

[ConstanceBtd]

Thanks for your answer.
Sorry for the mix-up, I did change the variable name afterwards and forgot to copy the new error message (same as the one I pasted but with clusters instead of cluster_matrix).

[digitalwright]

You need to specify myDistMatrix=clusters.

[ConstanceBtd]

Thank you, it works now !

[ConstanceBtd]

@digitalwright, I'm sorry to bother you again but I have another question regarding ASV clustering with DECIPHER.
I am using the code I found here to transform the ASV phyloseq that I already have into a phyloseq object clustered into OTUs:

d <- DECIPHER::DistanceMatrix(refseq(physeq_asv),type = "matrix", includeTerminalGaps = T, processors=1)

clusters <- DECIPHER::TreeLine(
  myDistMatrix=d, 
  method = "single",
  cutoff = 0.03, # use `cutoff = 0.03` for a 97% OTU 
  processors = 1)

## Use dplyr to merge the columns of the seqtab matrix for ASVs in the same OTU
# prep by adding sequences to the `clusters` data frame
clusters <- clusters %>%
  add_column(sequence = asv_sequences)

merged_seqtab <- seqtab %>% 
  t %>%
  rowsum(clusters$cluster) %>%
  t

With this code I encounter the following error:

> clusters <- clusters %>%
+   add_column(sequence = asv_sequences)
Error:
! The `.data` argument of `add_column()` must be a data frame as of tibble 2.1.1.
Backtrace:
 1. clusters %>% add_column(sequence = asv_sequences)
 2. tibble::add_column(., sequence = asv_sequences)
 3. lifecycle::deprecate_stop("2.1.1", "add_column(.data = 'must be a data frame')")
 4. lifecycle:::deprecate_stop0(msg)

The problem comes from the fact that I have updated the code with the TreeLine function (replacing the IdClusters one) but the output for TreeLine is a dendrogram which I have trouble transforming into a matrix to make the add_column work. Do you have any idea on how I could make this code work ?

Thanks again !

[digitalwright]

You need to add TreeLine(..., type="clusters").

[ConstanceBtd]

Thank you so much ! It worked