evaluate-binning: line 300: $4: unbound variable

[PhilPalmer]

Hi,
I am trying to benchmark a taxonomic binning pipeline.
I am first trying to get the binning-evaluation to work with 1st CAMI Challenge Dataset 1 CAMI_low. I have followed the simple instructions, having built the docker container philpalmer/binning-evaluation and created the relavant input, output and cache folders.

The input folder contains the following:

The contents of the biobox.yaml file are:

version: 0.10.0
arguments:
  sequences:
    path: /bbx/mnt/input/CAMI_low_RL_S001__insert_270_GoldStandardAssembly.fasta
    id: fas
    type: contig
    format: bioboxes.org:/fasta
  labels:
    path: /bbx/mnt/input/gsa_mapping.binning
    id: lab
    type: binning
    format: bioboxes.org:/binning/binning:0.9/taxbinning
  predictions:
    path: /bbx/mnt/input/gsa_mapping.binning
    id: STRING
    type: binning
    format: bioboxes.org:/binning/binning:0.9/taxbinning
  databases:
    taxonomy:
      path: /bbx/mnt/input/taxonomy
      id: taxonomy
      type: ncbi
      format: bioboxes.org:/taxonomy_ncbi_dumps

However on running of the container with the command: docker run -v $PWD/input:/bbx/mnt/input:ro -v $PWD/output:/bbx/mnt/output:rw -v $PWD/cache:/bbx/mnt/cache:rw philpalmer/binning-evaluation
I get the following error /opt/cami-evaluate-binning/evaluate-binning: line 300: $4: unbound variable

It seems as though while the binning and fasta files specified in biobox.yaml can be found some requried varaibles are not being passed to the script, specifically the prefix:

  gold_bin_file=$1
  query_bin_file=$2
  seq_len_file=$3
  prefix=$4
  btype=${5:-taxon}

1. Any idea why this is occurng and how it can be resolved?
2. I have manually generated the seq_len_file, shall I try calling the script inside the container passing the above varaibles as command line arguments?
3. What should the value of prefix be?
4. Should I be using AMBER instead for benchmarking a taxonomic binning pipeline?

Any help would be much appreciated, many thanks in advance

[fungs]

Hi Phil,

sorry for letting you wait, I just came back from vacation. Indeed, this biobox container represents an earlier (and hacked) version of measures used by CAMI. It should be all functional and correct. However, the measures where re-implemented in AMBER in an attempt to consolidate and join different methods. AMBER should also be easier to use with a single python interface, this earlier container is basically a collection of scripts put together.

The problem you are having seems to be due to a regression bug introduced by @AlphaSquad (2c691b7#diff-150af162eefa5234f9accdadbf012b30) by adding the line

 prefix=$4

outside of a shell function, which renders the container dysfunctional. You can try to check out an earlier revision or try AMBER instead. It is not likely that this repository will receive a fix soon. Sorry for any inconvenience!

Best,
Johannes

[PhilPalmer]

@fungs Okay, no problem, thanks for your help