1. 25 Aug, 2017 1 commit
  2. 07 Jul, 2017 2 commits
    • Mikaël Salson's avatar
      algo/: Count number of sequences independently of input format · ba7c8b9e
      Mikaël Salson authored and Mathieu Giraud's avatar Mathieu Giraud committed
      Use OnlineBioReader rather than a Fasta reader
      ba7c8b9e
    • Mikaël Salson's avatar
      algo/: Refactor Fasta to BioReader · 3d8ff242
      Mikaël Salson authored and Mathieu Giraud's avatar Mathieu Giraud committed
      We now have an abstract class to deal with biological sequence files. This
      will allow to more easily manage different file types.
      
      This commit only reorganizes the code so that we will be able to add a BAM
      reader easily. Functionnally the code should be equivalent to its previous
      version.
      
      Some functions that were not used have been removed.
      
      The operator>> has been removed as it was only used in unit testing. This
      operator is not convenient as having the filename may be useful to reopen the
      file or to know its extension, to guess the filetype.
      
      See #2016
      3d8ff242
  3. 14 Mar, 2017 3 commits
  4. 13 Mar, 2017 4 commits
  5. 07 Mar, 2017 2 commits
  6. 03 Mar, 2017 2 commits
  7. 19 Feb, 2017 1 commit
  8. 14 Feb, 2017 1 commit
  9. 02 Feb, 2017 3 commits
  10. 01 Feb, 2017 2 commits
  11. 30 Jan, 2017 5 commits
  12. 13 Jan, 2017 5 commits
  13. 05 Jan, 2017 1 commit
  14. 01 Dec, 2016 1 commit
  15. 07 Oct, 2016 1 commit
    • Mathieu Giraud's avatar
      vidjil.cpp: FineSegmenter in -c clones also considers e-value · 8f7219b8
      Mathieu Giraud authored
      Follows what is done for D in e6ffb915, and what was already done for V/J in -c segment.
      Even if it may be counter-intuitive, the results of segmentation thus actually depends on the number of reads:
      '-e 1' should mean that we expect one false positive segmentation in the output.
      Even if some e-values slightly change, this commit does not affect the results of the tests.
      
      We should investigate what happens when a FineSegmenter fails here.
      Moreover, is it sane to have the same '-e' paramater for fixing both e-values of Kmer and FineSegmenter?
      Or should we rather have another paramater?
      8f7219b8
  16. 28 Sep, 2016 1 commit
  17. 27 Sep, 2016 3 commits
  18. 26 Sep, 2016 1 commit
  19. 23 Sep, 2016 1 commit
    • Mikaël Salson's avatar
      vidjil.cpp (tests and doc): -t options defaults again to 0 · cd292aa0
      Mikaël Salson authored
      This reverts bf2f96fe and d317535b.
      
      After more than one year experience with -t 100 we realised that trimming leads to:
      – missing special cases where the V gene is heavily trimmed
      – giving bad unsegmentation causes: UNSEG only J appeared in cases where it didn't make sense
        those unsegmentation causes were replaced by UNSEG only V when changing -t to 0 (that also let us think that we could miss some V genes)
      
      However -t 0 does have a drawback that should be addressed. It is more likely to have
      spurious hits of V genes in the N-region (particularly with VDJ recombinations
      where the N-region is longer). Those spurious hits tend to shift the window towards
      the J gene.
      
      In the tests the modifications are due either to slight changes in number of windows
      (a small increase due to the previous reason) or to an increase in unsegmentation
      with really large windows (since windows are shifted we may not have
      enough space left to put a window).
      cd292aa0