low number of bins

[Valentin-Bio]

Hello developer I ran binny as following:

binny \
  --assembly /home/vberrios/comau/reads/assemblies/all_contigs/all_guys/unique_contigs.fasta  \
  --bam /home/vberrios/comau/reads/bam_files/sorted_bam/*.bam \
  --outputdir /home/vberrios/comau/remastered_mags/binny_bins \
  --threads 40 \
  --tmp_dir /home/vberrios/comau/remastered_mags/binny_tmp \

The contigs were generated with MEGAHIT, so I kept only the first strings before the first space. After that, I ran binny specifying the bam files (not a depth file).

I have assembled the reads individually (sample per sample for each one of my 93 samples) and stored all the contigs in a single FASTA file, I dereplicated the contigs at 100% of identity and mapped the reads against this set of contigs to generate 93 bam files. After that and after running Binny I only got 19 bins. I guess this can be related to some specification on the default config file provided by conda as I used it as default.

By consideration, I have used other binning tools and I was getting between 800 and 400 bins with a size > 2MB, so what can be happening here ? if it is worth to mention, the total number of contigs in the unique_contigs.fasta file is of ~ 1.2 million of contigs

binny.log

bests,

Valentín.

[ohickl]

Hi Valentin,

sorry for the late reply. binny does not have a multi-samples binning mode as e.g. semibin or vamb have, which sounds like what you are attempting. you'd have to either co-assemble all samples and then map each samples read sets against the single assembly, or you run binny per sample, so one individual assembly and read coverage file.
also, keep in mind that binny be default will filter out low quality bins, so the raw output will naturally have fewer bins than most other binners.

best

Oskar