[mira_talk] Re: bug

Hi Laurent,
did you use the sff_extract third party script? I had exactly the same error for some of my SFFs and it turned out that the sff_extract script produced wrong data. 
Once upon a time there was the command
sffinfo mysff.sff |sffinfo2mirafiles.tcl -project mysff
and applying this on the same sff worked good.

sffinfo comes with all the software from 454.

cheers
Burkhard


Laurent MANCHON schrieb:

Bastien Chevreux a écrit :

On Montag 04 Mai 2009 Laurent MANCHON wrote:

it returns:
Warning: "Read FUJ9VHQ01DRU97: tried to set 245 qualities although the
read has 242 bases.
"
->Thrown: void Read::setQuality(vector<base_quality_t> & quals)
->Caught: void ReadPool::loadDataFromFASTA(const string & filename,
const string & qualfilename, const bool generatefilenames, const uint8
seqtype)
..../....
Loaded 434802 reads, 345575 of which have quality accounted for.


Hello Laurent,
as Lionel already pointed out: that's no bug, but one of MIRAs safeguards which try to prevent you shooting yourself in the foot. There is no option to ignore bad input. And there will never be (sorry for that). MIRA is ... well I am pretty strict with these kind of things: either the input meets the expectations and can be assembled, or it does not. In such cases, one should really, really have a look at why the input is bad. In your case, ~90k reads (1/5th of the data) are corrupt and that's too much to simply ignore it.
I cannot even recommend to assemble without quality file and fake the default quality (-AS:bdq): what if some parts that should have been clipped were in effect not? And consist of adaptors? In my eyes that's too high of a risk.
Again, I'm sorry to say: please check the upstream data generation, eliminate the bugs there and then use good data to assemble :-) 

Regards,
  Bastien



thanks for this help bastien.





--
You have received this mail because you are subscribed to the mira_talk mailing 
list. For information on how to subscribe or unsubscribe, please visit 
http://www.chevreux.org/mira_mailinglists.html

Other related posts: