Binchicken Single - No MAGs

[mpdoane2]

Hi Sam,

I've run binchicken several times now, altering different parts of the script to see if I can get this to work. Unfortunately, I am not getting any results. My output says 17 samples had no targets with sufficient combined coverage for coassembly prediction. Log file attached here.

This particular run used this command:
binchicken single --forward-list $BGFS/forward_unzipped_list.txt --reverse-list $BGFS/reverse_unzipped_list.txt --cores 64 --output $OUTPUT --run-aviary

Output from the end of the log file:
[Fri Feb 14 17:31:34 2025]
Finished job 0.
25 of 25 steps (100%) done
Complete log: .snakemake/log/2025-02-14T165746.221711.snakemake.log
02/14/2025 05:31:34 PM WARNING: 17 samples had no targets with sufficient combined coverage for coassembly prediction
02/14/2025 05:31:34 PM WARNING: These are recorded at /cluster/jobs/doan0033/2508397/out_single_num2/coassemble/target/unused_samples.tsv
02/14/2025 05:31:34 PM INFO: Bin Chicken coassemble complete.
02/14/2025 05:31:34 PM INFO: Cluster summary at /cluster/jobs/doan0033/2508397/out_single_num2/coassemble/summary.tsv
02/14/2025 05:31:34 PM INFO: More details at /cluster/jobs/doan0033/2508397/out_single_num2/coassemble/target/elusive_clusters.tsv
02/14/2025 05:31:34 PM INFO: Aviary outputs at /cluster/jobs/doan0033/2508397/out_single_num2/coassemble/coassemble

This was a dataset with 17 samples. However, I have run this over a dataset with 85 samples and have returned no MAGs either. Using VAMB on each of these datasets, I was able to get >50 MAGs for the 17-sample dataset and > 100 for the 85-sample [dataset.](

bin_single-2508397_err.txt

)

Is there something I am not setting appropriately?

Thank you in advance,
Mike

[AroneyS]

Hi Mike,

"17 samples had no targets with sufficient combined coverage for coassembly prediction" means that there were no maker gene sequences with estimated coverage >10x. Can you let me know what is in your otu_table.tsv outputs (e.g. out_single_num2/coassemble/pipe/SAGCFN_23_00214_S1_read.otu_table.tsv)? Are they empty?

Also what are the input files (e.g. gz_files_single/SAGCFN_23_00228_S1_R1_001.fasta)? Bin Chicken is designed to work with fastq sequencing files as input.

Sam

[mpdoane2]

Hi Sam,

There is output in the files generated here. Here is one line which, if I am interpreting this correctly, would suggest 168.21x coverage.

gene sample sequence num_hits coverage taxonomy
S3.1.ribosomal_protein_L2_rplB SAGCFN_22_01149_S3_R1_001 GAATTGAAGCCTGGCAAAGGCGGACAATTAGTTCGTTCTGCAGGAACTTCAGCACAGGTT 102 168.21 Root; d__Bacteria; p__Bacillota; c__Bacilli; o__Bacillales_B; f__DSM-18226; g__Mesobacillus

Okay, great to know. I adjusted this as one of the errors I was troubleshooting before said DIAMOND only took .fasta files. I will adjust this back to .fastq input.

Thank you
Michael

[mpdoane2]

Hey Sam,

I have adjusted the code to use .fastq files. I am still getting the same result as above however. Is there anything else I should be checking?

Thanks
Mike

[AroneyS]

Where are the 17 samples from? Similar datasets? If each sample has plenty of hits, then it means that there aren't any overlaps between samples.
I'd have to see the otu tables to help beyond that.

[mpdoane2]

I have samples from two environments in this dataset. These are microbiome samples from the skin surface of a shark species and the water column from which they were sampled. The samples are indeed different. I was under the impression the tool found the most similar samples and then grouped those to retrieve the genomes.

Can you explain more what you mean by 'plenty' of hits? Maybe I am confused about how the tool functions, but is it attempting to identify and isolate MAGs? I assume the output will give fasta files of binned reads for each MAG?

I'm happy to share an OTU table if you want to look at it. If so, how many do you want to see and what is the best way to share these?

Thank you
Michael

[AroneyS]

Hi Mike,

Bin Chicken relies on their being matches of marker genes between samples. If there are no matches, then there are likely no shared species between your samples. Without shared species, there isn't any point in using those samples for differential abundance binning (or coassembly).

By "plenty", I just meant that as long as there are lots of marker gene sequences then SingleM probably ran correctly, so the issue is likely later on.

Sam