MCRazerToDo

1. For Herrn Kleffe:
   • script to parallelize by reads with normal version
   • don't do all reads at a time: go over contig with some read
     • script around?
     • given size of memory you want to use calculate read numbers
   • better expain qgram index
2. open addressing index
   • implement
   • documentation
   • tests
   • describe
   • comparison with "normal" index
     • speed (overview)
     • memory: how to mesure that?
   • what happen with q > 15?
     • how to make razers use it?
3. window find
   • implement
   • documentation
   • tests
   • describe
   • compare with "normal" version can't see a difference in speed
4. parallelization by reads
   • implement
   • split hits instead of going over the window multiple times
     • simple version of the pipline approach
     • would save memory since the bucket map of the index would not be doubled
     • problem: filtering would not be parallel anymore
   • documentation
   • describe
   • compare with script (see first point)
   • statistics
     • how many blocks per thread over position
       • output
       • analyse
     • waiting times
       • output
       • analyse
5. handle repeats
   • what kind of repeats
     • repeatfinder
     • analyse drosophila genome
   • how do they slow razers down
     • stats over filtration
     • stats over verification
6. parallelization with pipeline
   • …

1. Wrapper for different parallelization modes
   • none
   • by chromosom
   • by reads
   • pipeline