To: bochet at bobby
From: "RYBICKI, ED" <ED at molbiol.uct.ac.za>
Date: Fri, 2 Sep 1994 09:39:29 SAST-2
Subject: Re: picture in pileup
Reply-to: ed at molbiol.uct.ac.za
Priority: normal
X-mailer: WinPMail v1.0 (R1)
> When I run PILEUP (on VMS) and when it does not produce the
alignmemt because
> it has to introduce "too many gaps", the picture of the tree is
lost as well.
> In my understanding of the algorithm, this picture is produced
just after the
> pairwise comparisons, that is before the proper alignment starts.
Does anybody
> know of a way to keep this picture even when the alignment fails?
Yes - don't use PileUp...! Use CLUSTALV: in my experience this is
better for alignments where what you're trying to align are much the
same length as each other, AND it gives you a dendrogram in tabular
form from the initial pairwise alignment scores which you can
use to draw a real tree (PileUp just calculates a pretty-looking
UPGMA tree from the same pre-multiply aligned info). HOWEVER,
Clustal goes further in that you can then calculate a
neighbour-joining tree from MULTIPLY aligned sequence pairwise
comparisons (a whole lot more accurate - all the gaps are in the
same places in each sequence relative to any other). You should be
able to get it for VAX and for MAC and for PC
at the EMBL site.
_________________________________________________________________
_________________________________________________________________
| Ed Rybicki, PhD | Well, I tip my hat |
| (ed at micro.uct.ac.za) | To the new constitution |
| Dept Microbiology | Take a bow for the new revolution... |
| University of Cape Town | Then I get on my knees and pray |
| Private Bag, Rondebosch | We don't get get fooled again... |
| 7700, South Africa | |
| fax: xx27-21-650 4023 | - Pete Townshend, 1972 |
| tel: xx27-21-650 3265 | (Won't get fooled again) |
------------------------------------------------------------------
