mikewolfe / bacterial_reseq Goto Github PK

Add rule to rename_breseq_output and update snakefile to know about it

Change line 32 in quality_control.smk from
out = lookup_sample_metadata(sample, "file_path", pep)
to
out = lookup_sample_metadata(sample, "infile_path", pep)

"infile_path" is the name of the column in the sample sheet, not "file_path"

Although breseq is a great tool, it is not designed to deal with large changes containing exogenous DNA not found in the reference. To quote the breseq methods paper (https://doi.org/10.1007/978-1-4939-0554-6_12):

The re-sequencing strategy employed by breseq breaks down when there is sufficient sequence divergence between the sample and the reference genome to lead to mismapping of reads, such as when there is a high level of local nucleotide divergence or there are novel sequences in the sample that are not in the reference genome. In these cases, it may be valuable to use de novo assembly tools, such as Velvet (10.1101/gr.074492.107) or ALLPATHS-LG (10.1101/gr.141515.112), on either the unmapped reads after a breseq run (located at data/unmatched.*) or all reads and then compare assemblies to the reference genome or to one another using tools for comparing whole-genome sequences, such as MUMmer (10.1186/gb-2004-5-2-r12).

I think we should follow their suggestion and use these tools to perform a de-novo assembly of the sequencing samples.

• Decide on which de-novo assembler to use. Are these the best for bacterial genomes and short read sequencing? Is there something newer that is better?
• Check to make sure any tool is conda installable (i.e. MUMmer is https://anaconda.org/bioconda/mummer)
• Create a new module in workflow/rules/denovo_assembly.smk to run the assembly
• Add rules to compare the assembled genome against the reference genome. MUMmer is likely the best for this.
• Decide what set of output files need to be given to the end user
• Define how a user of the pipeline will indicate they want de-novo assembly ran. This could be a required column to the metadata sheet that tells the pipeline whether to run assembly or not on a sample

Consumers would like a BAM file to scrutinize for each sample. General parameters for alignment should be fine.