rvicedomini
commented
3 years ago Dear Ilnam,
First of all thanks for trying Strainberry with your data.
From the log you provided I don't see anything unusual. Straiberry seems able to identify two strains. This would mean that either there are only two strains collapsed by metaFlye in your input contig or there are more strains which have a very high sequence identity and Strainberry was not able to separate them. It is also possible to obtain a fragmented assembly in output and, due to the higher error ratio and the simple scaffolding procedure, I expect it to happen more frequently with Nanopore data than Pacbio.
I would also like to mention that you are testing Strainberry with a bacteriophage culture, which is not the use case of our tool (which was tested only with bacteria). In any case I don't think I can tell you more than this without having a look at the input data (the fasta and bam files you used).
If you can let me have access to such data I could have a deeper look to assess if there are some anomalies/bugs related to the tool. In this case, you can directly contact me at riccardo.vicedomini [at] pasteur.fr
To have a better understanding of what is happening, I would also suggest:
- to perform some polishing on your input data before providing it to Strainberry (e.g., one run of Medaka)
- to provide the tabular version of your BLAST hits (that would be great)
- to share, if possible, the following files generated by Strainberry in the output directory:
strainberry_n2/10-separation/vcfs/contig_3.vcf- the
*.1coordsfiles in the directorystrainberry_n2/20-assembly/trimmed strainberry_n2/assembly.scaffolds.info.tsv- all files in
strainberry_n2/30-scaffolds
If you are willing to share anything above, you can contact me at the same e-mail address metioned above :) Except for the vcf file (which contains SNV nucleotides/positions), all other files do not contain any nucleotide sequence.
Best, Riccardo
ilnamkang
gitraffica
Hi,
I'm trying to apply strainberry to my nanopore data obtained from the sequencing of a bacteriophage culture.
metaFlye generated a single contig of ~80 kbp from the nanopore data. But, based on read mapping and the SPAdes assembly of short reads (and hybrid), I'm sure that multiple highly similar strains exist in the culture. That's why I'm trying to apply strainberry to this data.
Strainberry seemed to work fine, but the seven sequences in assembly.scaffolds.fa are shorter than the input sequence, ranging from 667 bp to 65,084 bp. I attach a graphic summary of Blastn between the input (query) and the strainberry output sequences.
Below is the log of strainberry.
[2021-06-15 22:19:59] Starting Strainberry v1.1-a90d3b3 [2021-06-15 22:19:59] ### performing 2-strain separation [2021-06-15 22:19:59] SNP calling and phasing [2021-06-15 22:34:57] average Hamming ratio improved to 0.1926 [2021-06-15 22:34:57] separating reads [2021-06-15 22:41:00] assembling strain haplotypes [2021-06-16 06:10:36] scaffolding [2021-06-16 06:10:59] mapping reads to strain-separated scaffolds [2021-06-16 06:11:16] ### performing 3-strain separation [2021-06-16 06:11:16] SNP calling and phasing [2021-06-16 06:18:34] average Hamming ratio did not improve enough: 0.2098 [2021-06-16 06:18:34] Output strain-separated assembly available at: contig3.Sberry/assembly.scaffolds.fa [2021-06-16 06:18:34] Strainberry finished successfully
# Assembly statistics:
- Input assembly [contig3.fasta] Sequences: 1 Length: 80321 bp N50: 80321 bp
- Final iteration (2-strain separation) [contig3.Sberry/assembly.scaffolds.fa] Sequences: 7 Length: 151290 bp N50: 30447 bp
I'm wondering whether this result is usual or not. If not, how would I be able to handle the problem?
Thanks.
Ilnam