search

wanpinglee / MOSAIK

reference-guided aligner for next-generation sequencing technologies
http://code.google.com/p/mosaik-aligner/
55 stars 19 forks source link

Segmentation fault when attempting to write jumping file for large references #5

Open chartl opened 10 years ago

chartl commented 10 years ago

This is identical to

https://code.google.com/p/mosaik-aligner/issues/detail?id=101#makechanges

and is still an open issue.

(pasted for convenience)

What steps will reproduce the problem?

  1. Fasta file with human genome (build GRCh37) only chromosomes + MT, in human.fasta
  2. Build colorspace reference using: "MosaikBuild -cs -fr human.fasta -oa human.dat" (works)
  3. Make jump database using: "MosaikJump -ia human.cs.dat -mem 60 -hs 24 -out human.cs.24.jump"

What is the expected output? What do you see instead?

A jump database

What version of the product are you using? On what operating system?

Segmentation fault

Please provide any additional information below.

I have tried several combinations of hash size and memory usage with MosaikJump, however when the hash size gets to big (bigger than 14) the it results in a segmentation fault in the step 'writing jump positions database' at 0%. (So building and sorting the hash works, writing to file fails)