Open ctxchris opened 7 years ago
Hi Chris,
Thanks for reporting this!
The --diploid issue is a silly bug, the option is currently not even used in tulipbulb.perl, apologies for that.
The other issue may have to do with the script trying to read the SAM file and the PacBio reads simultaneously, i.e. trying to find the read matching the alignment. For this, both files need to be more or less in the same order (by read), which will happen when using BWA. The 'more or less' allows for some local shuffling due to alignment multithreading (BWA not reporting alignments in exactly the order they appear in the reads FASTA). By default, we read 5000 lines ahead to catch this (line 22, $fbuffermax). I can reproduce your error by setting $fbuffermax really low (5) and using 4 alignment threads. This needs a better fix!
If you want to try playing with this, I recommend using the --nobundles setting, tulipbulb then won't open thousands of files but just one for the scaffold sequences.
Alternatively, perhaps tulipbulb does not recognize your FASTA headers at all, and proceeds right to the end of the reads file trying to find a header? Could you perhaps share what your PacBio FASTA headers look like, and how they end up in the SAM file?
Also, not sure which genome or how much data you used, but those tulipseed stats worry me: very many scaffolds, and each consists of just a few seeds. We got thousands of seeds per scaffold.
Chris
Are you reversing your index and query in bwa mem? According to docs, you align your pacbio reads to your illumina-based seeds, and not the other way around.
Hi,
I did the following:
tulipseed.perl seems to have run successfully. The last lines say:
Running tulipbulb.perl produced the following warnings/errors:
Even though only --diploid is passed
Several thousand time I see these:
The corresponding lines in the seeds file illumina.fasta seem always to be header lines:
In the pacbio data they correspond to sequences:
In my working folder there are several thousand files named like this:
And they all only contain the pipe symbol:
Do you have an idea what could have gone wrong?
Thank you Chris