The final subworkflow that needs converting to nf-...
# team-maintainers
s
The final subworkflow that needs converting to nf-test,
fasta_index_dna
, needs a decision making about it, that ties into the use of bwa/mem more generally. This subworkflow can make an index for any of five different aligner tools (bwa/mem, bwamem2, snapaligner, bowtie, dragmap). It's inputs are the fasta file, which tool you want to make an index for, and an
altliftover
file. Bowtie, snapaligner and dragmap are all fine, but the two bwa modules are not currently working with how the subworkflow is currently coded for the
altliftover
file. If used (which makes bwa/mem understand
_alt
chromosomes properly), this file needs to sit inside the folder that is generated by bwa/index. There are three options how to incorporate this as far as I see it: 1. Add an extra process in the subworkflow to add the alt file into the bwa folder ◦ makes the subworkflow work, but doesn't make it particularly obvious that it is happening/available. 2. Add an extra input to bwa index, to take the alt file and just pass it through into the bwa output folder 3. Add an extra input to bwa mem to use the alt file as a separate channel (staging it correctly with the other index files) I suspect the best option is to make it part of bwa index personally Some previous discussion at https://nfcore.slack.com/archives/C04PPJLT1T3/p1748248543711989
3️⃣ 6
l
<!subteam^S044DU0NG68> What should we do therefore for this one ?
l
I like two, or 1 but with a message at then about what happened, maybe?
a
I also prefer option 2 3 Makes the bwa/~index~*mem* module responsible for its own required files (which I understand can be optional in some cases). I think it fits the guidelines of modules should define all required
path
inputs. (edit: i incorrectly typed the modules' names)
m
I like 3 or 2
j
Definitely 3
(just got back home hope meeting was good!)
r
bwa index doesn't do anything with the file right? Just stages it in and out? In that case 3
party this 3
m
That's an issue similar to bismark
s
Ok, think I can do that then. Anyone have any experience with actually creating .alt liftover files? It'd be nice to have a test file suitable for sarscov2.
m
AFAIK, liftover files are made available by the publisher of the genome
and the main usecase is GRCh38, which actually has alt contigs
s
Sure, but we don't generally test with GRCh38
m
In some cases we do
most GATK tools are tested with GRCh38 either with chr 21 or 22
s
is it the whole chromosome?
m
I think just a slice
s
I have a feeling one of them is completely fake, in that it is a region of the genome that is actually all Ns
ok, chr21 should probably be usable...
m
I know that in sarek we do use both depending on the data
maxime 1
s
Sarek is using it
s
I'm a bit confused by the chr21.fasta file in the test-datasets. The initial PR suggested it was from GRCh38, but it seems to have a different number of characters to the one I have....
shakefist 2
a
I've been making a pipeline that slices genomes to actually small regions and not entire chromosomes but I've been a bit lax at keeping it documented: https://github.com/adamrtalbot/nf-bore/pull/1
Ah never mind, it won't help you make an alt index
When I last did this I'm pretty sure I did some artisanal hand crafted alt index, because my "genome" was small enough to create an alt contig for 1 region
the alt index is basically a SAM file
but juussstt different enough to be annoying
shakefist 1
🤣 1
m
Of course it is