Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo

Close

#89 Making gap5 db inconsistent state via cutoffs/join

open
nobody
Gap5 (15)
5
2014-08-26
2011-03-30
Bastien Chevreux
No

Hi James,

attached is a valid small CAF (3 contigs) and its broken sibling after passing
through gap5. The broken one has a "shift" of 20 bases between reads and
consensus as the reads start covering the contig only at position 20.

Here are the steps to recreate:

1. tg_index -C mira_out.caf
2. gap5 mira_out.0.g5d
3. View -> "Find internal joins" -> "OK"
4. In the contig comparator, simply take any forward/forward match and invoke
join edtor. This example works with the longest match (the one between the
first and second contig)
5. In the joind editor, scroll to the left where one contig begins overlapping
with the other one (see provided screenshot).
6. In the join editor: for the contig that just starts and by using the "<"
key, move the cutoff of reads in a way that the contig starts, say, 20
bases later. (see provided screenshot)
7. Click "Join"
8. In the "Join and quit Editor" requester, click "No" (!)
9. File -> "Export sequences" -> (as CAF)
10. Look at the resulting file (which I called broken): the "Assembled_from"
lines are these:

--------
Sequence : mira_rep_c1
Is_contig
Padded
Assembled_from 230375b.t2 290 20 60 330
Assembled_from 230375b.t1 317 20 31 328
Assembled_from 230375c.t3 97 419 60 382
Assembled_from 230375b.t3 102 310 64 272
Assembled_from 122533c.t3 71 299 31 259
Assembled_from 122533a.t3 72 491 31 450
Assembled_from 122533b.t3 72 444 31 403
Assembled_from 230375a.t3 84 498 34 448
Assembled_from U13e07f02.t2 335 848 31 544
Assembled_from U13e07d02.t2 1413 651 126 888
Assembled_from U13a03g02.t1 697 1413 44 760
--------

Note that no reads cover positions 1 to 19. This is (I think) also an invalid
CAF.

Bonus question: now that I have such a database with a couple of these invalid
contigs, any idea how I can fix that? :-)

Best,
Bastien

Discussion

  • Archive with files to recreate error

     
    Attachments
    • labels: --> Gap5