alignment questions...pooled-seq, multiple references, pseudogenes, plastid genomes

[jullee]

Thank you mbblack and Brian Bushnell for the replies!

Unfortunately, I think my collaborator is quite set on working with BWA for the alignment and he will be doing that part of the pipeline. It looks like I will be coming in afterwards and be in the position of needing to extract mtDNA and cpDNA information from the resulting SAM or BAM file (e.g. the goal would be to partition the full SAM file into three new separate files for the nuclear, mtDNA and cpDNA data). I think this may be pretty straightforward using standard unix commands, but this will be my first time working with these types of files so if anyone has any additional thoughts, I'd be interested in hearing them...

Thanks!

[Brian Bushnell]

1) Is it standard to always include the mitochondria and chloroplast genome as part of the reference genome or do people usually only align reads to the nuclear genome?

Yes, they should all be together when mapping for greatest accuracy.

2) Do we have to set any of the parameters in BWA differently if we have multiple reference genomes in the same fasta file (e.g. the nuclear, chloroplast and mitochondrial genome)?

No. Most references will have multiple scaffolds anyway; aligners don't care if they are different chromosomes, different organelles, or different organisms.

3) Is it straightforward to separate the results for the nuclear and plastid genomes downstream (e.g. is it that by indexing the reference genome, we will be able to somehow partition the SAM/BAM file into data for our analysis of just nuclear SNP frequencies and data that goes into our analysis of the plastid genomes)?

Well... it is if you use BBSplit.

bbsplit.sh ref=plant.fa,mito.fa,chloroplast.fa in=reads.fastq basename=out_%.sam outu=unmapped.fastq -Xmx29g

"-Xmx29g" should be adjusted to the amount of RAM the computer has, roughly 85% of the total. This will align to all of the references at once, but create multiple output files:
out_plant.sam (reads mapped best to the plant)
out_mito.sam (reads mapped best to mito)
out_chloroplast.sam (reads mapped best to chloroplast)
unmapped.fastq (reads that did not map)

If you have paired reads, you can use "in1=read1.fq in2=read2.fq" for input.

4) Finally, wondering if anyone has a sense of how common plastid pseudogenes are in plant nuclear genomes? My thinking with the alignment of our reads to all three genomes is (in part) that we will be able to detect and avoid these types of pseudogenes but how important is this?

Pseudogenes are not conserved so I wouldn't worry about them too much, if you align to all references at once, since they will have SNPs that make the pseudogene reads go to the pseudogenes and the real gene reads go to the real genes. If you align to the references separately it would be more problematic.

[mbblack]

I cannot help with the plant specific questions. But, for mapping, I usually have my reference genome, and then a filter file to exclude things mapping to mtDNA and non-protein coding genomic regions. So anything getting mapped to the filter file contents just gets excluded from the genomic mapping altogether.

Of course, I don't care about mitochondrial genes nor non-coding genes. And that sort of filter mapping is very simply to set up in LifeScope.

In your case, it depends what you'd like to do with such genes. If you want to analyze them independently downstream, then using a filter reference may be best (you can save the results of the filter reference mapping separatly from the genomic mapping). Basically get your mtDNA and chloroplast mappings separate from your genomic mappings all at the same time that way.

Otherwise you can include everything into one reference genome file and map to that. It just depends on what you intend to do with it all later on, and which will give you the most straightforward file or set of files.

[jullee]

Hi Everyone,

I am working with pooled-seq data from several plant populations. From this data we are primarily interested in obtaining nuclear SNP frequencies by population but would also like to recover whatever information we can about the chloroplast and mitochondrial genome. I have the following questions:

1) Is it standard to always include the mitochondria and chloroplast genome as part of the reference genome or do people usually only align reads to the nuclear genome?

2) Do we have to set any of the parameters in BWA differently if we have multiple reference genomes in the same fasta file (e.g. the nuclear, chloroplast and mitochondrial genome)?

3) Is it straightforward to separate the results for the nuclear and plastid genomes downstream (e.g. is it that by indexing the reference genome, we will be able to somehow partition the SAM/BAM file into data for our analysis of just nuclear SNP frequencies and data that goes into our analysis of the plastid genomes)?

4) Finally, wondering if anyone has a sense of how common plastid pseudogenes are in plant nuclear genomes? My thinking with the alignment of our reads to all three genomes is (in part) that we will be able to detect and avoid these types of pseudogenes but how important is this?

Thank you in advance for the help! Sorry if some of these are naive questions: still new to all of this!