No mapping of isoforms

[Ingeborg3]

Hi,

Thank you for your help with the GTF file!
I realized that I cannot use FastMap in this case, because it only allows a single mismatch between peptide sequence and exon, and I have isoform sequences from multiple accessions of the same species, but only a reference genome for one of the accessions. I.e., there is both variation within species due to alternative splicing and amino acid substitutions.

However, when I try to run the default BLAT+Spaln2, none of the isoforms are mapped. I work with the vicia faba genome, which is ~13Gb and has 6 chromosomes. Therefore, the genome coordinates are very large. E.g., the end position of chromosome 2 is 1.554 Gb. Could this cause any problems?
Also, in the original Spaln2, the default is to only accept canonical dinucleotide pairs at the ends of introns (the -ya flag is set to N=0 by default). Is this also the case when Spaln2 is run through mirage2? This could maybe also cause problems.

Lines like the following are printed:
Use of uninitialized value $b in numeric lt (<) at /work/Mirage/build/Bureaucracy.pm line 93.
Use of uninitialized value $b in numeric lt (<) at /work/Mirage/build/Bureaucracy.pm line 93.
Use of uninitialized value $RangeEnds[0] in subtraction (-) at /work/Mirage/build/Quilter2.pl line 4276.
Use of uninitialized value $RangeEnds[1] in subtraction (-) at /work/Mirage/build/Quilter2.pl line 4276.
Use of uninitialized value $b in numeric lt (<) at /work/Mirage/build/Bureaucracy.pm line 93.
Use of uninitialized value $b in numeric lt (<) at /work/Mirage/build/Bureaucracy.pm line 93.
Use of uninitialized value $RangeEnds[0] in subtraction (-) at /work/Mirage/build/Quilter2.pl line 4276.
Use of uninitialized value $RangeEnds[1] in subtraction (-) at /work/Mirage/build/Quilter2.pl line 4276.
Use of uninitialized value $b in numeric lt (<) at /work/Mirage/build/Bureaucracy.pm line 93.
Use of uninitialized value $b in numeric lt (<) at /work/Mirage/build/Bureaucracy.pm line 93.
Use of uninitialized value $RangeEnds[0] in subtraction (-) at /work/Mirage/build/Quilter2.pl line 4276.
Use of uninitialized value $RangeEnds[1] in subtraction (-) at /work/Mirage/build/Quilter2.pl line 4276.
Use of uninitialized value $ChrEnds[0] in subtraction (-) at /work/Mirage/build/Quilter2.pl line 4088.
Use of uninitialized value $ChrEnds[1] in subtraction (-) at /work/Mirage/build/Quilter2.pl line 4088.

Do you know how to fix this?

Best,
Ingeborg

[alexander-nord]

Hi Ingeborg,

Thanks for bringing these issues to our attention!

I'll try to respond satisfactorily to each of your questions, but I'm right at the most hectic part of finishing up grading for our semester, so apologies if some of my responses seem overly brief -- I'll be able to dig into more thorough debugging in a week or so.

It sounds like in your case you're absolutely right to lean on the BLAT+Spaln route through the pipeline. I haven't personally tested Mirage on genomes with very large chromosomes, so it's entirely possible that's at the route of the issue -- maybe one of the tools runs into formatting issues when reporting large coordinate values? In any event, I'll try to look into that soon (and absolutely add some vicia faba tests to my debugging tests!).

We run Spaln2 with the following parameters:
-Q3 -O1 -S1 -ya3 -yz4 -yy4
This should allow for flexibility in the dinucleotide signatures of splice sites.

It looks like the error outputs you've captured highlight an edge case in Mirage's parsing of Spaln outputs -- thanks for finding this! To bore you with the details, there's an intentional undercount of Spaln hits that appears to be automatically adjusted without considering whether an "undercount" of 0 hits is actually accurate (in cases where there are no viable hits). That should be an easy patch to push through soon.

Hopefully that hits on all the key questions you raised, but don't hesitate to let me know if I missed anything.

Best,

Alex

[Ingeborg3]

Hi Alex,

Thank you for your reply.
And thank you for wanting to dig into debugging - it is very helpful. Please say if you want me to send some example files when you get to it. And no stress btw, it sounds intense with grading.

And great with the Spaln2 parameters.

Best,
Ingeborg