Rosella can be run multiple different ways in order to make using it as easy as possible.
The most likely situation is that you are running rosella in conjunction with other
binning algorithms, like metabat2. If so, then you likely already have coverage values precomputed
using coverm contig using -m metabat.
To perform mag recovery:
rosella recover -r scaffolds.fasta --coverage-values coverm.cov -o rosella_bin/ -t 24
If you have yet to run CoverM then rosella can generate the coverage values for you!
This is especially useful if you have both long and short reads as they can be passed
to rosella in tandem. Don't be afraid to pass multiple samples at once to rosella either,
it can handle it and keep everything in order.
rosella recover -r scaffolds.fasta -1 short_s[12345].1.fastq.gz -2 short_s[12345].2.fastq.gz --longreads nanopore.fastq.gz -o rosella_bins/
You can keep the BAM files that are created by rosella by including the --bam-file-cache-directory
flag. Once the coverage values are calculated, they will be stored in the output direcotry along with
the kmer frequency file.
Rosella can also be used to refine the results of other binning algorithms. The required input for this process is: - The original assembly FASTA file - MAGs from Rosella or another binning algorithm - Coverage values for the original assembly OR a set of reads to calculate them with Optionally, you can also provide the results of CheckMv1 or CheckMv2 (or AMBER) to limit refinement to only MAGs that are contaminated above the max contmiantion threshold.
rosella refine -r scaffolds.fasta -d metabat_bins/ -x fna -C coverm.cov -o refined_bins/ -t 24
OR
rosella refine -r scaffolds.fasta -d metabat_bins/ -x fna -1 short_s[12345].1.fastq.gz -2 short_s[12345].2.fastq.gz --longreads nanopore.fastq.gz -o refined_bins/ -t 24
The output of this process will be two folders, refined_bins/ and unchanged_bins/. The former will contain the refined MAGs and the latter will contain the MAGs that were not refined (either because they were below the contamination threshold or no change occurred after refinement).
The main output for rosella will be a set of MAGs denoted rosella_bin_X.fna. How many bins you get depends on your
samples. Additionally, the kmer frequency table will be present: kmer_frequencies.tsv. And the coverage values if
they were calculated by rosella: coverage.tsv. Finally, you'll get a pretty UMAP projection plot coloured
by potential MAG clusters. This plot isn't necessary but it helps you interpret how well rosella partitioned out your contigs.
If you see only a couple of big noisy clusters then maybe something went wrong and you'll want to fiddle with a few of
the UMAP parameters. This is unlikely though, but if you feel like you do need to then please feel free to raise an issue
on this GitHub and I'll answer your question and add my respone to the FAQ to help other users.
Powered by Doctave