From bryanlepore at mail.utexas.edu Thu Nov 1 18:34:45 2007 From: bryanlepore at mail.utexas.edu (Bryan W. Lepore) Date: Thu Nov 1 18:34:51 2007 Subject: [Discuss] avestruc : glycines with C-beta? Message-ID: hi observation : i get C-betas in the .pdb from avestruc where glycines are conserved. must be doing something wrong...? - doing something like: avestruc -polyA -f filename_scan.3 -ident -cons -o output.dat are -ident and -cons mutually exclusive? am i missing something? thanks -bryan p.s: just a thought : it'd be cool if avestrc could write out C-gammas, C-deltas, if they are conserved. From bryanlepore at mail.utexas.edu Thu Nov 1 21:39:45 2007 From: bryanlepore at mail.utexas.edu (Bryan W. Lepore) Date: Thu Nov 1 21:39:56 2007 Subject: [Discuss] dstamp (v4.3) Message-ID: i sometimes get segfaults w/ dstamp, without any error messages. i compared some .clean files that 'work' w/ the 'bad' ones, and it looks like there just isn't enough equivalenced regions, or too many different residues. is this correct? i think i can get around it by editing the .clean by hand. i also see #define MAX_STAMP_NUM 10 #define MAX_N_SEQ 1000 but that's just the number of sequences, right? -bryan From tom at compbio.dundee.ac.uk Fri Nov 2 10:07:44 2007 From: tom at compbio.dundee.ac.uk (Tom Walsh) Date: Fri Nov 2 10:11:27 2007 Subject: [Discuss] avestruc : glycines with C-beta? In-Reply-To: References: Message-ID: <472AF6F0.8010106@compbio.dundee.ac.uk> Bryan W. Lepore wrote: > observation : i get C-betas in the .pdb from avestruc where glycines > are conserved. This is because AVESTRUC builds 'ghost' CBs for glycines when you use the -polyA option to build a polyalanine structure. > are -ident and -cons mutually exclusive? No, you get both if you use both. -ident takes precedence over -cons so positions where sequence identity is 100% will be labelled with the name of the conserved amino acid rather than whatever property is conserved. Tom Walsh From tom at compbio.dundee.ac.uk Fri Nov 2 10:36:45 2007 From: tom at compbio.dundee.ac.uk (Tom Walsh) Date: Fri Nov 2 10:38:49 2007 Subject: [Discuss] dstamp (v4.3) In-Reply-To: References: Message-ID: <472AFDBD.6010202@compbio.dundee.ac.uk> Bryan W. Lepore wrote: > i sometimes get segfaults w/ dstamp, without any error messages. i > compared some .clean files that 'work' w/ the 'bad' ones, and it looks > like there just isn't enough equivalenced regions, or too many > different residues. Can you send me an example of a file that doesn't work? I'll try running DSTAMP in the debugger to see if I can figure out what's going on. BTW you should consider using Jalview (www.jalview.org) as an alternative to DSTAMP + Alscript if you want to generate figures of STAMP alignments. Tom From geoff at compbio.dundee.ac.uk Fri Nov 2 10:47:09 2007 From: geoff at compbio.dundee.ac.uk (Geoff Barton) Date: Fri Nov 2 10:47:23 2007 Subject: [Discuss] dstamp (v4.3) In-Reply-To: <472AFDBD.6010202@compbio.dundee.ac.uk> References: <472AFDBD.6010202@compbio.dundee.ac.uk> Message-ID: > BTW you should consider using Jalview (www.jalview.org) as an alternative > to DSTAMP + Alscript if you want to generate figures of STAMP alignments. Good idea except Jalview won't automatically box the structurally conserved regions etc, so you would have to highlight these by hand in Jalview. Geoff. From tom at compbio.dundee.ac.uk Fri Nov 2 18:05:27 2007 From: tom at compbio.dundee.ac.uk (Tom Walsh) Date: Fri Nov 2 18:07:15 2007 Subject: [Discuss] dstamp (v4.3) In-Reply-To: References: <472AFDBD.6010202@compbio.dundee.ac.uk> Message-ID: <472B66E7.8090400@compbio.dundee.ac.uk> Hi Bryan, I think I have a fix for this. Can you replace dstamp.c in your copy of the STAMP source with the attached file and rerun the BUILD script? You should then have a version of DSTAMP that works. Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: dstamp.c Type: text/x-csrc Size: 18305 bytes Desc: not available Url : http://www.compbio.dundee.ac.uk/pipermail/discuss/attachments/20071102/a527caa3/dstamp-0001.bin From bryanlepore at mail.utexas.edu Fri Nov 2 18:43:24 2007 From: bryanlepore at mail.utexas.edu (Bryan W. Lepore) Date: Fri Nov 2 18:43:29 2007 Subject: [Discuss] dstamp (v4.3) In-Reply-To: <472B66E7.8090400@compbio.dundee.ac.uk> References: <472AFDBD.6010202@compbio.dundee.ac.uk> <472B66E7.8090400@compbio.dundee.ac.uk> Message-ID: On Fri, 2 Nov 2007, Tom Walsh wrote: > I think I have a fix for this. Can you replace dstamp.c in your copy > of the STAMP source with the attached file and rerun the BUILD script? yes ( gcc34 (GCC) 3.4.6 20060404 (Red Hat 3.4.6-4) ) .. well, i adjusted makefile (from makefile.linux) to use gcc34, and yes. > You should then have a version of DSTAMP that works. not only that, looks like all the other stamp executables compiled perfectly also. thanks a lot! -bryan From bryanlepore at mail.utexas.edu Fri Nov 2 21:33:58 2007 From: bryanlepore at mail.utexas.edu (Bryan W. Lepore) Date: Fri Nov 2 21:34:09 2007 Subject: [Discuss] multalign w/ stamp .bloc (or .blc) and one .seq : no alignment? Message-ID: i've been having troubles aligning a sequence w/ stamp output. i can get the single sequence to line up along side the stamp output, but close inspection shows that particular sequence motifs are not lining up. in fact there are no gaps at all in the sequence - shouldn't the sequence get gapped to line up nice w/ the stamp output? i've been reading/working off of this section of the manual : http://www.compbio.dundee.ac.uk/manuals/amps/section3_9.html#SECTION0009000000000000000 ... since it and the examples seem most pertinent. i think i tried most of the obvious tricks, including using 'iterations=int', editing the .seq to look like myoglobins.seq, checking bash_mult_vg.com, ... i got this to work a long time ago, but it was by cutting the sequence up a lot - not really the right way to go. are some sequences just too poorly conserved to align? these are < 19% ID. any hints, suggestions much appreciated. -bryan From geoff at compbio.dundee.ac.uk Mon Nov 5 09:29:41 2007 From: geoff at compbio.dundee.ac.uk (Geoff Barton) Date: Mon Nov 5 09:45:50 2007 Subject: [Discuss] multalign w/ stamp .bloc (or .blc) and one .seq : no alignment? In-Reply-To: References: Message-ID: Hi Bryan, I think you have answered the question really. If the sequence identity is low, then the alignment will be difficult to do by any method. If you are certain that the motifs should align, then you can force the alignment in a number of ways, but it is probably best just to edit the alignment in a good interactive editor such as jalview (www.jalview.org). There are lots of factors that can affect whether known motifs in a sequence align as you would expect. Overall sequence identity, difference in lengths of sequences, range of similarity in the set of sequences youare aligning, etc. You can get an idea of how alignable a set of sequences are by doing pairwise comparisons with AMPS and randomisations and then clustering the data. This is explained in the manual and in various reviews I've written. If the pairwise S.D. scores (Z-scores) are below about 6 then the alignment will have a lot of errors in it. Above 6 most of the core elements should be correct, but there will still be errors. Although I am fond of AMPS since I wrote it, it is now 20 years old and has not been developed at all since it was first written. It is competitive in accuracy to clustalW and other methods from the 1990s, but more recent methods such as T-Coffee, PROBCONS and MAFFT out-perform it on many alignment problems. You might also like to try 3D-coffee from Des Higgin's group. This allows you to combine seqence and strutural information in a single multiple alignment in an automatic way. I've not used it, but the idea is a good one and the alignments I have seen done with it look good. Hope that helps, Geoff. On Fri, 2 Nov 2007, Bryan W. Lepore wrote: > i've been having troubles aligning a sequence w/ stamp output. i can get the > single sequence to line up along side the stamp output, but close inspection > shows that particular sequence motifs are not lining up. in fact there are > no gaps at all in the sequence - shouldn't the sequence get gapped to line up > nice w/ the stamp output? > > i've been reading/working off of this section of the manual : > > http://www.compbio.dundee.ac.uk/manuals/amps/section3_9.html#SECTION0009000000000000000 > > ... since it and the examples seem most pertinent. i think i tried most of > the obvious tricks, including using 'iterations=int', editing the .seq to > look like myoglobins.seq, checking bash_mult_vg.com, ... i got this to work a > long time ago, but it was by cutting the sequence up a lot - not really the > right way to go. > > are some sequences just too poorly conserved to align? these are < 19% ID. > any hints, suggestions much appreciated. > > -bryan > _______________________________________________ > Discuss mailing list > Discuss@compbio.dundee.ac.uk > http://www.compbio.dundee.ac.uk/mailman/listinfo/discuss > -- Geoff Barton, Professor of Bioinformatics, School of Life Sciences University of Dundee, Scotland, UK. geoff@compbio.dundee.ac.uk Tel:+44 1382 385860/388731 (Fax:385764) www.compbio.dundee.ac.uk