[mira_talk] Re: Problem with Paired 454, MP Illumina, MIRA and Bambus

  • From: Robin Kramer <kodream@xxxxxxxxx>
  • To: mira_talk@xxxxxxxxxxxxx
  • Date: Wed, 6 Jul 2011 16:14:39 -0600

I have had a very difficult time getting Bambus to scale to Illumina Paired
end scales.  I was determined enough to rewrite all of the perl conversion
scripts, simply to realize that the core c++ program gromit will crash very
unexpectedly, and if debug the C++ code you will find that the results are
very suspect, possible do to a non-scaling design, or maybe some
implementation flaw, and the project has been non-active for six years, and
there is no one answering the help messages for Bambus, though the AMOS
project is still active.

Sincerely yours,

Robin


2011/7/6 Nestor Zaburannyi <nestor@xxxxxxxxxxxxx>

> Dear all. Excuse me for a long letter.
>
> It seems i am following many steps of Juan specifically from this thread,
> but i am opening new one as suggested:
> //www.freelists.org/post/mira_talk/fn-in-pairedend-reads,4
>
>
> I have 454 paired data and Illumina Mate-Pair data. It took me some time to
> figure out that not the Bambus is wrong... Long story short:
>
> When i use all the data and both pairing schemes, MIRA result is
> outstanding, but Bambus, depending on several factors which i can present,
> either crashes or chokes for DAYS on contigs constructed by MIRA.
>
> These are ridiculous results from bambus when using both pairing schemes.
> Note the "203415" number.
>
> Library:        lib_roche_4kb
> no. valid links:        2124
> no. incorrect len. links:       281
> no. incorrect ori. links:       13
> no. unchecked links:    715
>
> Library:        lib_illumina_4kb
> no. valid links:        613
> no. incorrect len. links:       203415
> no. incorrect ori. links:       9
> no. unchecked links:    408
>
>
> Most of the contigs are not joined properly and we get hundreds of
> thousands "Invalid" links with consistent distance between them. Something
> like:
>
> contig_c46 (260871, 277262) ====> v:93 l:1382 o:0 ====> contig_c33 (277800,
> 308449)
> ...
> Invalid length:
>  library lib_illumina_4kb:
>    5:117:14385:3543/1   16063   16113 <---  ...  4435 ...  --->    3619
>  3669 5:117:14385:3543/2
>    5:33:7647:14602/1   15968   16018 <---  ...  4702 ...  --->    3791
>  3841 5:33:7647:14602/2
>    5:33:8278:18410/1   16134   16184 <---  ...  4736 ...  --->    3991
>  4041 5:33:8278:18410/2
>    5:47:1673:4256/1 15946   15996 <---  ...  4660 ...  --->    3727
>  3777 5:47:1673:4256/2
>    5:11:1902:7439/2   14782   14832 <---  ...  4721 ...  --->    2624
>  2663 5:11:1902:7439/1
>    5:35:18689:20790/1   14758   14808 <---  ...  4323 ...  --->    2202
>  2244 5:35:18689:20790/2
>    5:9:8606:11124/1   15673   15723 <---  ...  5317 ...  --->    4111
>  4161 5:9:8606:11124/2
>    5:50:8824:11116/2   15738   15788 <---  ...  4299 ...  --->    3158
>  3208 5:50:8824:11116/1
>    ...
>
> Now, i tried to assemble subset of the data in three combinations:
> 1- only PE information used
> 2- only MP information used
> 3- without any template information
>
> Results of MIRA assembly are absolutely comparable, but take a look at
> bambus result:
>
> 1:
>
> Library:        lib_roche_4kb
> no. valid links:        5746
> no. incorrect len. links:       984
> no. incorrect ori. links:       4
> no. unchecked links:    284
>
> Library:        lib_illumina_4kb
> no. valid links:        479
> no. incorrect len. links:       19346
> no. incorrect ori. links:       1
> no. unchecked links:    115
>
> 2:
>
> Library:        lib_roche_4kb
> no. valid links:        569
> no. incorrect len. links:       23
> no. incorrect ori. links:       16
> no. unchecked links:    87
>
> Library:        lib_illumina_4kb
> no. valid links:        5303
> no. incorrect len. links:       404
> no. incorrect ori. links:       21
> no. unchecked links:    231
>
> 3:
>
> Library:        lib_roche_4kb
> no. valid links:        632
> no. incorrect len. links:       13
> no. incorrect ori. links:       16
> no. unchecked links:    75
>
> Library:        lib_illumina_4kb
> no. valid links:        5785
> no. incorrect len. links:       274
> no. incorrect ori. links:       32
> no. unchecked links:    271
>
>
> As you can see, using 454 paired-end information for MIRA assembly results
> in the disaster downstream. Please, correct me if i am wrong.
>
>
> Sincerely yours
> Nestor
>
>
> --
> 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:

  1. Home
  2. mira_talk
  3. Archive
  4. 07-2011