[maker-devel] Help on error-Repeat masker
Carson Holt
carsonhh at gmail.com
Tue Apr 2 07:15:28 MDT 2013
The best evidence is from mRNAseq or ESTs of the same species. ESTs and
mRNAseq from related species can be used, but if protein annotations are
available use those instead. This is because cross species nucleotide
alinements must be translated in all 6 reading frames (3 for the query and 3
for the subject), which would basically make run times increase by 6 fold.
You can try giving the cross species alignments to maker as if they were
from the same species instead (est= option), fewer will align, but run times
will not be overwhelming. Then provide the protein annotations from the
related species combined with uniprot (maker can take comma separated lists
for the input files). You can use either the program CEGMA from Ian Korf's
lab or alternatively maker's protein2genome option to build an initial
annotation set to use for training. Then train SNAP, Augustus, and GeneMark
(Genemark self trains). For the last run let all 3 predictors run together
with protein2genome now turned off. Given that the genome is only 50Mb and
you have a lack of alignment evidence, you can probably safely set
keep_preds=1 on the second run as the false positive rate is usually quite
low for gene dense organisms and you won't get many annotations from maker
otherwise without more evidence alignments. Perform your first and second
runs in the same location, so maker can automatically reuse the same
alignments (the second run is always very fast this way as maker won't have
to rerun blast and exonerate).
If your organism is a fungi (I'm just guessing because of the small genome
size) you can also use this gene prediction parameter resource from Jason
Stajich
--> https://github.com/hyphaltip/fungi-gene-prediction-params
Thanks,
Carson
From: Hud Hud <hudarul at yahoo.com>
Reply-To: Hud Hud <hudarul at yahoo.com>
Date: Monday, 1 April, 2013 9:16 PM
To: Carson Holt <carsonhh at gmail.com>
Subject: Re: [maker-devel] Help on error-Repeat masker
Thanks, i now have better insight regard to the cpu cores.. i have other
questions...i dont have other info or evidences of my own genome, i only
have assembled contigs....recently JGI sequenced a species that closely
related to my genome (at genus level), and i have access to the data
(protein, est, rna-seq reads,transcript, gene models,gff)
1.I have run maker (MWAS) using diff set of evidences, such as protein and
est(JGI) and est(JGI) and uniprot database ..but both run produced diferent
no of predicted genes....so my question, what is the best evidences to be
used to support my annotation..is it more preferred to use larger dataset
such as uniprot rather than using the data from JGI (even it closely
related)
2. can i use rna-seq data (from JGI) to be used in maker...ive denovo
assembled the rnaseq using clc genomics.
Thanks
From: Carson Holt <carsonhh at gmail.com>
To: Hud Hud <hudarul at yahoo.com>
Sent: Tuesday, April 2, 2013 5:01 AM
Subject: Re: [maker-devel] Help on error-Repeat masker
That's not too bad It's best to choose a few large contigs (1-2Mb total) to
run with at first and then use those results to help configure the rest of
the run. For the final run you may want to consider splitting onto multiple
machines if your machine has limited cpu power. It will take you ~150
hours on 1 cpu core depending on the size of alignment datasets - ESTs and
proteins. More cpu cores will allow it to run faster (see graph below from
the MAKER2 paper). I imagine that your machine probably has at least 4 cpu
cores.
Most bioinformatics labs have multi cpu Linux boxes (I.e. 24-32 cpu cores),
some have clusters available to them (100's to 1000's of cpu cores), and a
few just launch maker on multiple lab desktop machines all writing to the
same network mounted output directory.
Thanks,
Carson
Thanks,
Carson
From: Hud Hud <hudarul at yahoo.com>
Reply-To: Hud Hud <hudarul at yahoo.com>
Date: Monday, 1 April, 2013 4:48 PM
To: Carson Holt <carsonhh at gmail.com>
Subject: Re: [maker-devel] Help on error-Repeat masker
Its about 50mb
From: Carson Holt <carsonhh at gmail.com>
To: Hud Hud <hudarul at yahoo.com>
Sent: Tuesday, April 2, 2013 4:44 AM
Subject: Re: [maker-devel] Help on error-Repeat masker
How big is the genome?
--Carson
From: Hud Hud <hudarul at yahoo.com>
Reply-To: Hud Hud <hudarul at yahoo.com>
Date: Monday, 1 April, 2013 4:37 PM
To: Carson Holt <carsonhh at gmail.com>
Subject: Re: [maker-devel] Help on error-Repeat masker
owh thanks so much,now i know whats going wrong, its the cygwin... i'll try
dual boot then as my genome over 10 mb..thanks for your time
From: Carson Holt <carsonhh at gmail.com>
To: Hud Hud <hudarul at yahoo.com>
Cc: "maker-devel at yandell-lab.org" <maker-devel at yandell-lab.org>
Sent: Tuesday, April 2, 2013 4:29 AM
Subject: Re: [maker-devel] Help on error-Repeat masker
I found it odd because perl.exe is a windows extension not used in Linux,
but it confirmed my suspicions. You can't use maker with cygwin. There are
several things that will break because it's not really Linux. You can use
Virtual Box instead to install a virtual Linux machine -->
https://www.virtualbox.org/. Alternatively you can try and dual boot your
system with a Linux partition.
Virtual Box will allow you to run maker on small datasets, depending on the
size of the genome you want to run maker with it may be fine. But I would
not recommend running anything over 10 megabases (it won't fail, it will
just take a very long time).
Thanks,
Carson
From: Hud Hud <hudarul at yahoo.com>
Reply-To: Hud Hud <hudarul at yahoo.com>
Date: Monday, 1 April, 2013 4:21 PM
To: Carson Holt <carsonhh at gmail.com>
Subject: Re: [maker-devel] Help on error-Repeat masker
1. owh its odd?im using windows8 but for maker im using cygwin
2. when i type which perl i get this /usr/bin/perl
3. when i type ./Build repeatmasker i got this
cygwin warning: MS-DOS style path detected: \Users\Dora Preferred POSIX
equivalent is: /cygdrive/c/Users/Dora CYGWIN environment variable option
"nodosfilewarning" turns off this warning. Consult the user's guide for more
details about POSIX paths:
http://cygwin.com/cygwin-ug-net/using.html#using-pathnames WARNING:
RepeatMasker was already found on this system. Do you still want MAKER to
install RepeatMasker for you?
is there any prob with this, or can i just proceed with the installation?
From: Carson Holt <carsonhh at gmail.com>
To: Hud Hud <hudarul at yahoo.com>
Cc: "maker-devel at yandell-lab.org" <maker-devel at yandell-lab.org>
Sent: Tuesday, April 2, 2013 3:59 AM
Subject: Re: [maker-devel] Help on error-Repeat masker
What kind of system (OS) are you running on? 'perl.exe' seems odd.
It appears that the perl is different for maker and RepeatMasker. What do
you get when you type 'which perl' on the command line?
I think you need to reinstall RepeatMasker at a minimum. To do that -->
> cd /home/maker-2.27-beta/maker/src
> ./Build repeatmasker
--Carson
From: Hud Hud <hudarul at yahoo.com>
Reply-To: Hud Hud <hudarul at yahoo.com>
Date: Monday, 1 April, 2013 3:53 PM
To: Carson Holt <carsonhh at gmail.com>
Subject: Re: [maker-devel] Help on error-Repeat masker
Thanks for the reply
1. Yes i set up the maker myself as own user but i dont know how to check
for the mounting things
2. Im calling maker directly and i've tried this cat
/home/maker-2.27-beta/maker/bin/maker | grep "#\!" and it gaves me this
#!/usr/bin/perl.exe
From: Carson Holt <carsonhh at gmail.com>
To: Hud Hud <hudarul at yahoo.com>; "maker-devel at yandell-lab.org"
<maker-devel at yandell-lab.org>
Sent: Monday, April 1, 2013 10:50 PM
Subject: Re: [maker-devel] Help on error-Repeat masker
This appears to be a permissions issue either for the /u1/local/bin/
directory or RepeatMasker setup. Did you set maker up yourself as your own
user or did someone else do it for you, perhaps as root? Is /u1/local/bin/
on an NFS mount.
If it's a mounting issue I found this via google the exact same issue-->
>> I needed to add the 'exec' option to the /etc/fstab file when mounting that
>> partition.
>> If it says 'defaults' on the line in /etc/fstab, then it also means you don't
>> have exec rights on it.
Are you using the same perl to run maker as you are using for RepeatMasker?
For example, are you calling perl directly and giving the path to maker or
are you calling maker directly and letting it use the version of perl it was
installed with. Try this to see which perl maker was installed with -->
cat /home/maker-2.27-beta/maker/bin/maker | grep "#\!"
You may have to have to reinstall RepeatMasker and possibly maker.
Thanks,
Carson
From: Hud Hud <hudarul at yahoo.com>
Reply-To: Hud Hud <hudarul at yahoo.com>
Date: Sunday, 31 March, 2013 4:02 PM
To: "maker-devel at yandell-lab.org" <maker-devel at yandell-lab.org>
Subject: [maker-devel] Help on error-Repeat masker
Hello, i have some problem when runnning maker, i've got this kind of error,
what could possibly go wrong here? Thnks so much
setting up GFF3 output and fasta chunks
doing repeat masking
running repeat masker.
#--------- command -------------#
Widget::RepeatMasker:
cd /tmp/maker_WOVHsi;
/home/maker-2.27-beta/maker/exe/RepeatMasker/RepeatMasker
/home/maker-2.27-beta/maker/data/contig.maker.output/contig_datastore/61/0D/
contig172//theVoid
.contig172/contig172.0.simple.rb -dir
/home/maker-2.27-beta/maker/data/contig.maker.output/contig_datastore/61/0D/
contig172//theVoid.contig172 -pa 1 - lib /tmp/maker_WOVHsi/b1piBcWHlH
#-------------------------------#
sh: /home/maker-2.27-beta/maker/exe/RepeatMasker/RepeatMasker:
/u1/local/bin/perl: bad interpreter: Permission denied
ERROR: RepeatMasker failed
--> rank=NA, hostname=Homis
ERROR: Failed while doing repeat masking
ERROR: Chunk failed at level:0, tier_type:1
FAILED CONTIG:contig172
ERROR: Chunk failed at level:2, tier_type:0
FAILED CONTIG:172
examining contents of the fasta file and run log
_______________________________________________ maker-devel mailing list
maker-devel at box290.bluehost.comhttp://box290.bluehost.com/mailman/listinfo/m
aker-devel_yandell-lab.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://yandell-lab.org/pipermail/maker-devel_yandell-lab.org/attachments/20130402/7c60323a/attachment-0003.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 08EB6777-DA72-45CA-8E05-07928457B9BE.png
Type: image/png
Size: 61806 bytes
Desc: not available
URL: <http://yandell-lab.org/pipermail/maker-devel_yandell-lab.org/attachments/20130402/7c60323a/attachment-0003.png>
More information about the maker-devel
mailing list