[mira_talk] Re: Saving a crashed assembly

On Mittwoch 17 Juni 2009 Jan van Haarst wrote:
> I have an assembly that got killed.

Yikes ... let me guess: a eukaryote?

> Is there a way to not have to start from the start ?

No, sorry, there is no way to restart the assembly when it went down. I've had 
only very few requests for that and implementing this feature is actually 
somewhat down on the priority list. Sorry again.

Depending on when it crashed, there may be intermediate results that you can 
re-use though. Your assembly was killed at a stage where the CAF file for a 
given pass is present in it's entirety in the log directory (if "-OUT:otc=yes" 
is used, which is currently standard for I think all --job combinations of 
MIRA). 

If you were killed after, e.g., pass 4 of 5, then the result for pass 4 
shouzld already be pretty good and you could think to use that. Look for 
"SRMc" tags in the file. If only very few are present, then most repeats will 
have been resolved and you can think of using that file as result.

You could also re-launch MIRA with the temporary CAF as input file (then 
reducing the number of passes accordingly), though I think I remember that 
this is advisable only for bacterial projects because eukaryotes with thir 
repeats may cause even longer running time than before.


> The last messages were:
> Localtime: Wed Jun 10 13:01:05 2009
>[ ...]
> It started at
> Localtime: Sat May 30 22:20:10 2009
> You can imagine that I'm not looking forward to start from the start  ...

I can, indeed.

If everything fails: relaunch the assembly with -SK:mnr=yes:nrr=15 ... this 
should keep out the very worst repeats from the assembly and go pretty quick. 
Actually I'm thinking of putting this as standard for "--highlyrepetitive"

Hope this helps, please keep me informed if something does not work out as 
expected.

Regards,
  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: