Henrik,
yealing's second option is also how I do it. If you have local blast its not
that bad.....
Alex
________________________________
Van: artemis-users-boun...@sanger.ac.uk namens Yealing
Verzonden: di 11-5-2010 9:57
Aan: Henrik Lantz
CC: artemis-users@sanger.ac.uk
Onderwerp: Re: [Artemis-users] Multiple contigs and Blastall-results
Hello Hendrik
Not very sure about the blastall from blast+ package, but the blastall package
from NCBI blast package only outputs the gi/accession number for all the blast
hits in -m 8 (table) format.
We have two ways to deal with this:
(1) run blastall and use -m 0, which defaults and output all the long names,
then use a script to parse the file into a table format. The parsing can be
really slow, we're using a bioperl script. this is how most of the people in my
lab does it.
(2) run blastall and use -m 8, which outputs the table format with short names.
Then, using a script, collect all gi/accession the significant blast hits in a
text file, use batch entrez to get the long names, then use another script to
replace the short names in the blast output file with the long names. this is
how i do it, again, my scripting is really bad so I apologize for not sharing
the script.
All the best,
-- yealing --
On Tue, May 11, 2010 at 3:16 PM, Henrik Lantz <henrik.la...@mikrob.slu.se>
wrote:
Hi
A big thank you to Tim, Yealing and Alex for all the help!
Here is what I have done so far: I got help from a friend to write a
perl script that transforms the coordinates in the blast result file to
cumulative coordinates. This file can then be read as a feature in Artemis just
fine and maps perfectly onto the contigs.
I also made some progress with the alternative discussed with Tim,
i.e., loading the contig file in Artemis and choosing "write all bases in FASTA
format" and then blasting this file. I managed to avoid the memory related
problems I had with blastall by using the blast+ package and the legacyscript
included there. The blast results can then be read as a feature in Artemis
without any problems.
Of these two approaches I prefer the first one since this allows me to
keep the original contigs.
The "problem" I am having now is getting the names of the proteins into
Artemis. The blast results only include the systematic name of the blast hits,
not the full names in understandable English. I am sure I am just missing
something very simple here, and will soon have a solution, but if anyone has a
suggestion I would also be very interested in hearing about it. Still in the
learning phase...
Cheers,
Henrik
________________________________________
Från: artemis-users-boun...@sanger.ac.uk
[artemis-users-boun...@sanger.ac.uk] för Yealing
[yealingt+arte...@gmail.com <mailto:yealingt%2barte...@gmail.com> ]
Skickat: den 29 april 2010 11:59
Till: Tim Carver
Kopia: artemis-users@sanger.ac.uk; Henrik Lantz
Ämne: Re: [Artemis-users] Multiple contigs and Blastall-results
Hi
I've had the same problem in the past. There sometimes are reasons why
we do not wish to join the contigs before we send for a blast, for example once
we join the contigs, the other contig informations are lost.
I've been able to map the blast results correctly in Artemis, but
because I'm not a very good scripter, things might work or break using the
bash-awk script that I have. So I usually end up doing it sort of manually for
a lot of different people with different data sets. Here's how I do it:
1, we need a file containing <contig name> in the first column and
<contig length> in the second column. The make or break problem is that in this
file, the contig names must be arrange to the exact order as Artemis puts them.
2, from there, I use a script to generate a file with <contig name> in
the first column and <cummulative contigs length at start of contig> in the
second column. This can be done with excel too, if you're not familiar with
scripting.
3, I then process the blastall result file, by modifying the start and
stop coordinates (adding the cummulative contigs length at start of contig to
the start and stop coordinate).
After that, (usually if nothing bad happens) you will be able to just
load the blastall result file in Artemis. Best thing is we get to keep all the
details that we might or might not need.
Cheers,
-- yealing --
On Thu, Apr 29, 2010 at 5:02 PM, Tim Carver
<t...@sanger.ac.uk<mailto:t...@sanger.ac.uk>> wrote:
Hi Henrik
You have found the correct solution. It is just that the blast reports
coordinates from the start of the sequence and so they need joining up
before you do the blast. I am not sure why you get that memory error.
If you
have access to another machine you may want to try the blast there.
Alternatively you could possibly try a smaller number of contigs and
then
use the EMBOSS application 'union' to join them back up.
Regards
Tim
On 4/29/10 8:09 AM, "Henrik Lantz"
<henrik.la...@mikrob.slu.se<mailto:henrik.la...@mikrob.slu.se>> wrote:
> I was hoping I could get some help with a newbie Artemis question.
Very new to
> all this.
>
> I have made a denovo assembly of a fungus using MIRA and 454-data
only. The
> resulting fasta file with around 4000 contigs loads into Artemis
fine. I can
> check all the contigs, find ORFs etc. The problem appears when I want
to
> import the results of a blastall search on the contig-datafile. All
> annotations from the blastall results are lumped into the first five
contigs,
> with the overwhelming majority in the first contig. Obviously not
correct. I
> am using the -m 8 flag for the blastall search. Looking through the
resultfile
> from blastall in a text editor I can see that the blastall search has
worked,
> and there are many interesting hits, but I would like to visualize
the results
> on the contigs.
>
> I read through the mail archive and found a user with a similar
problem
>
(http://www.mail-archive.com/artemis-users%40sanger.ac.uk/msg00463.html) and
> it seems one solution might be to save the contigs as a long
continuous file
> in Artemis, and then use that in the Blastall search. But when I try
that I
> get an error message from blastall:
>
> blastall(33748) malloc: *** mmap(size=1048576) failed (error code=12)
> *** error: can't allocate region
> *** set a breakpoint in malloc_error_break to debug
> Bus error
>
> I am running on MacOSX Snow Leopard with 20 GBs of memory.
> Any help to get an inexperienced user started would be very much
appreciated!
> /Henrik
> _______________________________________________
> Artemis-users mailing list
> Artemis-users@sanger.ac.uk<mailto:Artemis-users@sanger.ac.uk>
> http://lists.sanger.ac.uk/mailman/listinfo/artemis-users
_______________________________________________
Artemis-users mailing list
Artemis-users@sanger.ac.uk<mailto:Artemis-users@sanger.ac.uk>
http://lists.sanger.ac.uk/mailman/listinfo/artemis-users
_______________________________________________
Artemis-users mailing list
Artemis-users@sanger.ac.uk
http://lists.sanger.ac.uk/mailman/listinfo/artemis-users
