Add swarm-like tabular output to vsearch clustering options

[Patg13]

Hi,

Would it be possible to add an output option to vsearch clustering to produce a swarm-like tabular output ?

I'm referring to the swarm --output-file argument which produces a tabular file where each line is a cluster and contains all sequences composing that cluster.

I'm working on an amplicon pipeline and I want to give users the choice of using swarm or vsearch and since my original method is based on swarm and its output format, that would make things a lot easier if vearch could output the same format.

Thank you !

[frederic-mahe]

Hi @Patg13

Adding a new output format to vsearch would require substantial work (especially testing). Would you consider using a script to convert vsearch's --uc output to swarm's default output format?

Here is a bash function (convert_uc_to_swarm) that does that. I've added a sort_by_decreasing_size function in case you need the clusters to be sorted by size.

INPUT="a_fasta_file"

convert_uc_to_swarm() {
    # uc format: 10 column tab-separated table
    # swarm format: list of amplicon headers separated by spaces
    grep "^[HC]" | \
        sort -k2,2n -k1,1d | \
        awk '{if (NR == 1) {printf $9" " ; next}
              if ($1 == "C") {printf "\n"$9" "}
              else if ($1 == "H") {printf $9" "}
             }
             END {printf "\n"}' | \
                 sed 's/ $//'
}

sort_by_decreasing_size() {
    # sort by:
    #  - total abundance,
    #  - number of amplicons,
    #  - abundance of the seed amplicon
    awk 'BEGIN {FS = "[ =]" ; OFS = "\t"}
         {abundance = 0
          for (i=2 ; i<=NF ; i+=2) {abundance += $i}
          print abundance, NF / 2, $2, $0
         }' | \
             sort -k1,1nr -k2,2nr -k3,3nr | \
             cut -f 4
}

vsearch \
    --cluster_fast "${INPUT}" \
    --threads 4 \
    --id 0.97 \
    --quiet \
    --uc - | \
    convert_uc_to_swarm | \
    sort_by_decreasing_size

[Patg13]

Hello,
No problem at all, I was asking just in case it would be easy to implement on your side. I'll test your script in my workflow, thank you !