From owner-btk-mca@hgmp.mrc.ac.uk Fri Aug 10 18:53:38 2001 Return-Path: Received: from localhost (localhost [127.0.0.1]) by mercury.hgmp.mrc.ac.uk (Postfix) with ESMTP id BA29217A67 for ; Fri, 10 Aug 2001 18:53:37 +0100 (BST) Received: from localhost (localhost [127.0.0.1]) by mercury.hgmp.mrc.ac.uk (Postfix) with ESMTP id EF89717A6A; Fri, 10 Aug 2001 18:53:33 +0100 (BST) Received: from niobium.hgmp.mrc.ac.uk (niobium [193.62.192.41]) by mercury.hgmp.mrc.ac.uk (Postfix) with ESMTP id 21AF617A66; Fri, 10 Aug 2001 18:53:30 +0100 (BST) Received: (from news@localhost) by niobium.hgmp.mrc.ac.uk (8.9.3+Sun/8.8.8) id SAA07686; Fri, 10 Aug 2001 18:49:05 +0100 (BST) X-Authentication-Warning: niobium.hgmp.mrc.ac.uk: news set sender to using -f From: "ricnews" X-Newsgroups: bionet.molbio.methds-reagnts,bionet.metabolic-reg References: <7fcc7.34584$jA2.2726200@news20.bellglobal.com> <3B72AC27.684B541F@sheffield.ac.uk> Subject: Re: fitting with dynafit? Lines: 92 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4522.1200 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200 Message-ID: Date: Fri, 10 Aug 2001 13:49:36 -0400 X-Complaints-To: abuse@sympatico.ca Organization: Bell Sympatico To: btk-mca@hgmp.mrc.ac.uk.methods@hgmp.mrc.ac.uk Sender: owner-btk-mca@hgmp.mrc.ac.uk Precedence: bulk Well, when I started playing I decided I'd try something easy just to learn how to use the program before I tried fitting a more complex model to my data. So what I tried initially was fitting a self cleavage curve for a self-cleaving ribozyme. The simplest model to explain this reaction, which is what we typically use to derive the rate constant using the theoretical equation, is given by: Pre ==> P + D I then gave it some real experimental data (of good quality, covering the whole range of the curve) and a starting guess for the rate constant that was very reasonable. (Using non-linear fitting in excel or sigmaplot to fit to the theoretical equation for Pre(t) there is no problem getting a good fit using the same starting guess for the rate constant.) Much to my surprise, dynafit did miserably. It wasn't even close and visually its best fit is way off. I was only able to get it to converge to the correct fit by giving it as the start value the value obtained by fitting the same data in Excel. And even then, while it gave a good fit by eye, the fit had a lower R2 than the starting guess obtained through Excel... I also tried fitting a more complex equation to other data and it failed as well. But I'd like to be able to fit this simple model before I tackle anything harder. I'm just surprised it has such a hard time fitting such a simple model and was wondering if I'm doing something wrong in my script file or my options or set incorrectly somehow... You mentioned you had success fitting simple binding models so I don't quite know why I can't get equal success with this first order cleavage model... Thanks, Ricardo "Jim Reid" wrote in message news:3B72AC27.684B541F@sheffield.ac.uk... > Hi, > > I've cc'd this to bionet.matabolic-reg as it seems to be the closest to > a "numerical modeling in enzymology" newsgroup. > > Anyway I suppose it very much depends on what you are trying to do. > I've fitted binding data without too many problems but that isn't very > tricky. When I've tried to fit more demanding data sets (e.g. progress > curves at varied ligand and enzyme concs.) I've had more problems. On > the basis of _very_ little experience (you have been warned!) these can > be due to problems analogous to determining Vmax when [S] << Km. These > are easy to spot when an analytical equation describing the model is > available but harder when you are only dealing with progress curves. > Perhaps a good safety net would be to simulate curves for your fitted > parameters in concentration regions that are far from your > experimentally explored ones. This is a bit off topic now but might get > a discussion going. > > As for the initial value problem, I have suffered from this when trying > to simulate the first few minutes of the time course of an enzyme with a > long lag phase in an attempt to get a steady state rate. I am not > convinced that this is anything to do with the program (similar problems > with kin/fitsim), but that there are a lot of local minima and picking > good starting values is just essential. > > Looking back over this it doesn't look very helpful but I'd also be > interested in hearing about the success stories or problems that people > are having with these numerical methods. Are they worth the effort? > Always? Sometimes - It depends? Never, just design expts. to avoid > them? > > All the best, > > Jim > > ricnews wrote: > > > > Hi, > > > > I'm wondering if anyone successfully used dynafit to fit their data to a > > model? I tried but even with a simple test I couldn't get it to fit unless I > > gave it starting values that were basically the answer. Using starting > > values that converge easily in sigmaplot or excel dynafit still couldn't get > > the answer. > > > > I received no response from the author so I was hoping someone else might > > use the software and be able to help me out. > > > > Thanks, > > Ric .