From criopel at uoguelph.ca Wed Apr 15 17:35:33 2009 From: criopel at uoguelph.ca (Chad Riopel) Date: Wed, 15 Apr 2009 20:35:33 -0400 (EDT) Subject: [Mesquitelist] Restricting Parameters for Pagel Correlation Analysis in Mesquite? In-Reply-To: <1903164809.1733601239841956875.JavaMail.root@erie.cs.uoguelph.ca> Message-ID: <756114095.1734741239842133775.JavaMail.root@erie.cs.uoguelph.ca> Hello Everyone, Is it possible to restrict parameters within the Pagel94 correlation analysis module (like in Bayestraits) in order to test dependent models with less than 8 parameters? I'd appreciate any suggestions you might have. Thanks, Chad From peteremidford at yahoo.com Wed Apr 15 18:25:46 2009 From: peteremidford at yahoo.com (Peter E. Midford) Date: Wed, 15 Apr 2009 20:25:46 -0500 Subject: [Mesquitelist] Restricting Parameters for Pagel Correlation Analysis in Mesquite? In-Reply-To: <756114095.1734741239842133775.JavaMail.root@erie.cs.uoguelph.ca> References: <756114095.1734741239842133775.JavaMail.root@erie.cs.uoguelph.ca> Message-ID: Chad, Currently that's not possible, though it is on the list of things to do. Thanks for asking, Peter On Apr 15, 2009, at 19:35, Chad Riopel wrote: > Hello Everyone, > Is it possible to restrict parameters within the > Pagel94 correlation analysis module (like in Bayestraits) in order > to test dependent models with less than 8 parameters? I'd appreciate > any suggestions you might have. > > Thanks, > > Chad > _______________________________________________ > Mesquitelist mailing list > Mesquitelist at mesquiteproject.org > http://mesquiteproject.org/mailman/listinfo/mesquitelist -- Peter Midford Mesquite Developer From budd at embl-heidelberg.de Thu Apr 23 10:12:17 2009 From: budd at embl-heidelberg.de (Aidan Budd) Date: Thu, 23 Apr 2009 19:12:17 +0200 (CEST) Subject: [Mesquitelist] merging Taxa blocks Message-ID: Hi MESQUITEers I am trying to trace the evolution of some binary characters across trees using MESQUITE The tree is stored in NEWICK format, with the NEWICK string all on one line. The character data is stored in categorical simple format I load the tree into MESQUITE, and then include (File->Include File...) the character data into the same project. I can view the tree in a tree window, and the character matrix in a Character Matrix window, all working fine. The tree and the character matrix are associated with two different taxa blocks (which contain exactly the same list of taxon names) - and thus I can not trace the charcters over the tree. I **can** do this if I create a new character matrix "by hand", as I get the choice to associate the new character matrix with the taxa block I want. Any suggestions on how to integrate/remove one of/deal with these two taxa blocks so that I can trace the characters stored in one file over a tree stored in another, would be very much appreciated. Thanks Aidan -- ---------------------------------------------------------------------- Aidan Budd, PhD tel:+49 (0)6221 387 8530 EMBL - European Molecular Biology Laboratory fax:+49 (0)6221 387 8517 Meyerhofstr. 1, 69117 Heidelberg, Germany http://www.embl-heidelberg.de/~budd/ http://www-db.embl.de/jss/EmblGroupsHD/per_1807.html From budd at embl-heidelberg.de Fri Apr 24 00:50:31 2009 From: budd at embl-heidelberg.de (Aidan Budd) Date: Fri, 24 Apr 2009 09:50:31 +0200 (CEST) Subject: [Mesquitelist] merging Taxa blocks In-Reply-To: <4224e620904231618s5025e833g9c3fc452e08cb4cc@mail.gmail.com> Message-ID: Hi Daniel Indeed, you're right - if I load first the character data, and THEN the tree (it works both if I link or include the tree), the tree is automatically associated with the taxa block (and if I already have more than one taxa block, then I'm asked to choose which one to associated the tree taxa with) Thanks for the suggestion - I spent quite a while trying to that to work..! Basically - problem solved now! Would still be interested to hear if, having done it the "wrong" way around (i.e. loading tree first, then the characters), anyone does know a way of forcing the trees/characters to be associated with the same taxa block. Thanks again, Aidan On Fri, 24 Apr 2009, Daniel wrote: > Hi Aiden, > Without mesquite to hand for testing, I'd say try first opening the > character matrix in Mesquite. Then go to taxa&trees and try to get > your trees from the file - use the link option rather than include. > Maybe that'll work. > > good luck, > daniel > > 2009/4/23 Aidan Budd : > > Hi MESQUITEers > > > > I am trying to trace the evolution of some binary characters across trees > > using MESQUITE > > > > The tree is stored in NEWICK format, with the NEWICK string all on one > > line. > > > > The character data is stored in categorical simple format > > > > I load the tree into MESQUITE, and then include (File->Include File...) > > the character data into the same project. I can view the tree in a tree > > window, and the character matrix in a Character Matrix window, all working > > fine. > > > > The tree and the character matrix are associated with two different taxa > > blocks (which contain exactly the same list of taxon names) - and thus I > > can not trace the charcters over the tree. > > > > I **can** do this if I create a new character matrix "by hand", as I get > > the choice to associate the new character matrix with the taxa block I > > want. > > > > Any suggestions on how to integrate/remove one of/deal with these two taxa > > blocks so that I can trace the characters stored in one file over a tree > > stored in another, would be very much appreciated. > > > > Thanks > > > > Aidan > > > > -- > > ---------------------------------------------------------------------- > > Aidan Budd, PhD ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? tel:+49 (0)6221 387 8530 > > EMBL - European Molecular Biology Laboratory ?fax:+49 (0)6221 387 8517 > > Meyerhofstr. 1, 69117 Heidelberg, Germany > > > > http://www.embl-heidelberg.de/~budd/ > > http://www-db.embl.de/jss/EmblGroupsHD/per_1807.html > > > > _______________________________________________ > > Mesquitelist mailing list > > Mesquitelist at mesquiteproject.org > > http://mesquiteproject.org/mailman/listinfo/mesquitelist > > > -- ---------------------------------------------------------------------- Aidan Budd, PhD tel:+49 (0)6221 387 8530 EMBL - European Molecular Biology Laboratory fax:+49 (0)6221 387 8517 Meyerhofstr. 1, 69117 Heidelberg, Germany http://www.embl-heidelberg.de/~budd/ http://www-db.embl.de/jss/EmblGroupsHD/per_1807.html From budd at embl-heidelberg.de Fri Apr 24 03:28:38 2009 From: budd at embl-heidelberg.de (Aidan Budd) Date: Fri, 24 Apr 2009 12:28:38 +0200 (CEST) Subject: [Mesquitelist] Logging the results of analyses Message-ID: hi again, MESQUITEers I'm looking for a way to have MESQUITE write the **results** of its analyses to a text file Thanks to Daniel, I can now do the ML analysis of binary characters I was looking for in MESQUITE (comparing Mk1 and AsymmMk1 with LRTs). I'm doing this for 100 or so different tree/character datasets, and would like to keep a text record of the results of each analysis i.e. the likelihoods under the two different models, the results of correlation analyses etc. - I'll be browsing through each analysis using MESQUITE's GUI, but I'd also like to store information in a file to look at in a spreadsheet etc. (e.g. to compare the forward/backward rates of all ML AsymmMk1 models from all trees and characters - is forward always faster than backward? - questions like that) The MESQUITE log records the commands I issue, but I was hoping to find also a facility that logs also/only the **results** of analyses of this kind. I know I can look at the text representation of the trees, but this only describes the results of the most-recently done analysis (e.g. if I first check for correlation between characters 1 and 2, and then between characters 2 and 3, this window will only describe the log likelihood difference between chars 2 and 3) - I'd like to record the results of **all** these analyses. Any pointers to how to convince MESQUITE to do something like this much appreciated! Or perhaps this is simply something that MESQUITE isn't designed to be able to do and I should be trying out alternative software...? Thanks Aidan -- ---------------------------------------------------------------------- Aidan Budd, PhD tel:+49 (0)6221 387 8530 EMBL - European Molecular Biology Laboratory fax:+49 (0)6221 387 8517 Meyerhofstr. 1, 69117 Heidelberg, Germany http://www.embl-heidelberg.de/~budd/ http://www-db.embl.de/jss/EmblGroupsHD/per_1807.html From budd at embl-heidelberg.de Fri Apr 24 03:48:25 2009 From: budd at embl-heidelberg.de (Aidan Budd) Date: Fri, 24 Apr 2009 12:48:25 +0200 (CEST) Subject: [Mesquitelist] Logging the results of analyses In-Reply-To: <98EA056410DF47DF883C4DF894126A43@ebc.uu.local> Message-ID: Hi Alejandro Thanks for writing back so quickly In fact, that's not what I'm trying to do - rather, I have a set of phylogenies which describe relatioships between different sets of taxa (they are trees esimated from several different gene families) I'm looking, then, for trends in the evolutionary dynamics of features of the proteins, and (amongst other things) would be interested to see whether there are common trends in different types of characters across the different families While I'll want to look through the results of each family "manually", I'd also like to write the results in a way that doesn't involve copying-and-pasting the likelihoods and rate parameter values for each analysis into a spreadsheet, to make overal trends easier to spot - hence looking for a way to get the results of these analyses printed out from MESQUTE Perhaps I should just use Pagel's software directly, or use PAUP*, or something else, if I'm looking to generate those kinds of summaries (although i'd stick with using MESQUITE for the browsing of the data - just seems a shame to repeat the same thing in different software liket that...) Thanks again Aidan On Fri, 24 Apr 2009, Alejandro Gonzalez Voyer wrote: > Hello Aidan, > > I've missed previous posts so I'm not 100% certain if I understood your > question correctly. If what you want to do is compare an independent versus > a dependent model of evolution for two binary traits across a subsample of > likely phylogenies (taking phylogenetic uncertainty into account) then you > should instead use Bayestraits. In Bayestraits you can also compare rates of > "gains" versus "losses" of the two traits etc. > You could have a look at: Pagel and Meade 2006 Bayesian analysis of > correlated evolution of discrete characters by Reversible-Jump Markov Chain > Monte Carlo. Am Nat 167: 808-825. > > Cheers, > > Alejandro > > -----Original Message----- > From: mesquitelist-bounces at mesquiteproject.org > [mailto:mesquitelist-bounces at mesquiteproject.org] On Behalf Of Aidan Budd > Sent: den 24 april 2009 12:29 > To: mesquitelist at mesquiteproject.org > Subject: [Mesquitelist] Logging the results of analyses > > hi again, MESQUITEers > > I'm looking for a way to have MESQUITE write the **results** of its > analyses to a text file > > Thanks to Daniel, I can now do the ML analysis of binary characters I was > looking for in MESQUITE (comparing Mk1 and AsymmMk1 with LRTs). > > I'm doing this for 100 or so different tree/character datasets, and would > like to keep a text record of the results of each analysis i.e. the > likelihoods under the two different models, the results of correlation > analyses etc. - I'll be browsing through each analysis using MESQUITE's > GUI, but I'd also like to store information in a file to look at in a > spreadsheet etc. (e.g. to compare the forward/backward rates of all ML > AsymmMk1 models from all trees and characters - is forward always faster > than backward? - questions like that) > > The MESQUITE log records the commands I issue, but I was hoping to find > also a facility that logs also/only the **results** of analyses of this > kind. > > I know I can look at the text representation of the trees, but this only > describes the results of the most-recently done analysis (e.g. if I first > check for correlation between characters 1 and 2, and then between > characters 2 and 3, this window will only describe the log likelihood > difference between chars 2 and 3) - I'd like to record the results of > **all** these analyses. > > Any pointers to how to convince MESQUITE to do something like this much > appreciated! > > Or perhaps this is simply something that MESQUITE isn't designed to be > able to do and I should be trying out alternative software...? > > Thanks > > Aidan > > -- ---------------------------------------------------------------------- Aidan Budd, PhD tel:+49 (0)6221 387 8530 EMBL - European Molecular Biology Laboratory fax:+49 (0)6221 387 8517 Meyerhofstr. 1, 69117 Heidelberg, Germany http://www.embl-heidelberg.de/~budd/ http://www-db.embl.de/jss/EmblGroupsHD/per_1807.html From criopel at uoguelph.ca Fri Apr 24 07:05:52 2009 From: criopel at uoguelph.ca (Chad Riopel) Date: Fri, 24 Apr 2009 10:05:52 -0400 (EDT) Subject: [Mesquitelist] Pagel94 vs Discrete rate parameter calculation In-Reply-To: <1024835694.437071240581932949.JavaMail.root@erie.cs.uoguelph.ca> Message-ID: <1020253561.437281240581952357.JavaMail.root@erie.cs.uoguelph.ca> Hello, I'm sorry to bother you all again but I'm confused as to differences in results obtained between the Mesquite Pagel94 module and results obtained by Discrete for maximum likelihood analysis. The parameters for the 8 parameter dependent model based on Discrete are: q12 = 1.52188 q13 = 0.01739 q21 = 0.0005 q24 = 0.98443 q31 = 0.21346 q34 = 0.00061 q42 = 4.13408 q43 = 1.59419 and for the same data set from Mesquite I get: q12 = 2.0976873829078578E-4 q13 = 26.837141573633282 q21 = 0.0016498309403183997 q31 = 4.245508561704672 q24 = 12.481781495270134 q34 = 5.122817481113189E-6 q42 = 77.10900256167346 q43 = 1.3547783960645563 The one parameter that concerns me is q13. In Discrete it's one of the lowest parameters but in Mesquite it's one of the highest. This has huge implications for the biological interpretation of my data. I realize the manual states Mesquite calculates parameters with a different optimizer which explains the differences, however it's less clear on how to handle this situation. If anybody has experienced this issue and/or knows how to deal with it, I'd appreciate any feedback you might have. Thanks for your time, Chad From peteremidford at yahoo.com Fri Apr 24 07:48:53 2009 From: peteremidford at yahoo.com (Peter E. Midford) Date: Fri, 24 Apr 2009 09:48:53 -0500 Subject: [Mesquitelist] Pagel94 vs Discrete rate parameter calculation In-Reply-To: <1020253561.437281240581952357.JavaMail.root@erie.cs.uoguelph.ca> References: <1020253561.437281240581952357.JavaMail.root@erie.cs.uoguelph.ca> Message-ID: <27125398-1240-421F-9896-EDE9C16DFD65@yahoo.com> Chad, I agree the results seem to disagree. I set up the Pagel94 module to be more aggressive (search longer) and cover a larger range of parameters than Discrete does typically. The likelihood surface for the search tends to be difficult and small differences in the search implementation may give large differences in the final result. As I recall from experimenting with Discrete while implementing the Pagel94 module, Discrete seemed to limit parameter values to be less than 50 (though that was only an inference on my part), so the solution Mesquite found would never be found by Discrete. You might try tweaking the likelihood iterations in Mesquite and see how it affects your answer. The other question is how do the reported likelihoods compare? If we could bound the range of parameter searches, would people find it useful? Hope this helps a bit, Peter On Apr 24, 2009, at 9:05, Chad Riopel wrote: > Hello, > I'm sorry to bother you all again but I'm confused as to > differences in results obtained between the Mesquite Pagel94 module > and results obtained by Discrete for maximum likelihood analysis. > The parameters for the 8 parameter dependent model based on Discrete > are: > > q12 = 1.52188 > q13 = 0.01739 > q21 = 0.0005 > q24 = 0.98443 > q31 = 0.21346 > q34 = 0.00061 > q42 = 4.13408 > q43 = 1.59419 > > and for the same data set from Mesquite I get: > > q12 = 2.0976873829078578E-4 > q13 = 26.837141573633282 > q21 = 0.0016498309403183997 > q31 = 4.245508561704672 > q24 = 12.481781495270134 > q34 = 5.122817481113189E-6 > q42 = 77.10900256167346 > q43 = 1.3547783960645563 > > The one parameter that concerns me is q13. In Discrete it's one of > the lowest parameters but in Mesquite it's one of the highest. This > has huge implications for the biological interpretation of my data. > I realize the manual states Mesquite calculates parameters with a > different optimizer which explains the differences, however it's > less clear on how to handle this situation. If anybody has > experienced this issue and/or knows how to deal with it, I'd > appreciate any feedback you might have. > > Thanks for your time, > > Chad > _______________________________________________ > Mesquitelist mailing list > Mesquitelist at mesquiteproject.org > http://mesquiteproject.org/mailman/listinfo/mesquitelist -- Peter Midford Mesquite Developer From alejandro.gonzalezvoyer at ebc.uu.se Sat Apr 25 07:02:34 2009 From: alejandro.gonzalezvoyer at ebc.uu.se (Alejandro Gonzalez Voyer) Date: Sat, 25 Apr 2009 16:02:34 +0200 Subject: [Mesquitelist] Pagel94 vs Discrete rate parameter calculation In-Reply-To: <27125398-1240-421F-9896-EDE9C16DFD65@yahoo.com> References: <1020253561.437281240581952357.JavaMail.root@erie.cs.uoguelph.ca> <27125398-1240-421F-9896-EDE9C16DFD65@yahoo.com> Message-ID: <67F0E6A57E3F4FDD8F0046C2B52BFE13@ebc.uu.local> Hello, Somewhat surprised by the differences in the values of the transition parameters that are estimated by the two programs (Mesquite and Discrete) I ran a few tests. I compared the output of both programs for four different data sets, although all with the same phylogeny. In only one case did the outputs differ in the relative values given to the parameters, and not as drastically as in Chad's case. In all cases the difference in log likelihoods between the dependent and independent models differed very little between Mesquite and Discrete. I ran the tests both with traits which did not show signal of correlated evolution and for traits for which the dependent model differed significantly from the independent model. The values of the estimated parameters for the four tests I ran are the following in case they can be of interest: Test 1, here the dependent model was not superior to the independent model. Mesquite parameter values: q12 = 3.200236327797246E-4 q13 = 5.848048055299448 q21 = 17.74473468874265 q31 = 9.808628314405603 q24 = 10826.892738726705 q34 = 14.560492893238845 q42 = 3700.4464526316865 q43 = 2.820240126970863E-4 Discrete parameter values: q12 = 0,00199 q13 = 6,12172 q21 = 18,28947 q31 = 9,91644 q24 = 49,86896 q34 = 14,05269 q42 = 22,06118 q43 = 6E-5 Test 2 Again there is no evidence supporting correlated evolution. Mesquite parameter values: q12 = 11.09774854004778 q13 = 2.8865198877688444E-4 q21 = 34.34673365201564 q31 = 1.3229818724067253 q24 = 0.24691052663988952 q34 = 9.750697960038881 q42 = 0.0012542039667116233 q43 = 0.0012897813698686475 Discrete parameter values: q12 = 10,5867 q13 = 0,00368 q21 = 31,74877 q31 = 1,24965 q24 = 0,00663 q34 = 8,91272 q42 = 8E-5 q43 = 0,00113 Now with traits for which the dependent model differs significantly from the independent model. Test 3 This is the only case where the relative rank (in terms of whether it was the highest or lowest value) of the parameters differed between the two programs: Mesquite parameter values: Rank (from the highest value to the lowest) q12 = 0.6187577819139466 6 q13 = 27.674370166078745 5 q21 = 109.94731002993473 2 q31 = 77.32410520595383 3 q24 = 226.64200961651068 1 q34 = 33.048415713216784 4 q42 = 7.784907657338925E-4 7 q43 = 3.842601640028465E-4 8 Discrete parameter values: q12 = 0,00172 8 q13 = 20,10772 5 q21 = 45,38993 2 q31 = 50,03288 1 q24 = 25,6768 4 q34 = 40,74672 3 q42 = 0,00206 6 q43 = 0,00175 7 Test 4: Mesquite parameter values: q12 = 0.0026556759499011704 q13 = 11.654624023154076 q21 = 6.229025425726005 q31 = 0.005472588460444028 q24 = 31.53931649049757 q34 = 34.31480203662039 q42 = 10.582272292443719 q43 = 5.116190166503267E-5 Discrete parameter values: q12 = 0,00758 q13 = 11,3989 q21 = 0,0004 q31 = 0,00074 q24 = 46,85473 q34 = 34,73541 q42 = 12,82575 q43 = 0,00013 Cheers, Alejandro Alejandro Gonzalez Voyer Post-doc. Animal Ecology, Department of Ecology and Evolution Evolutionary Biology Centre (EBC) Uppsala University Norbyv?gen 18D 75236 Uppsala Sweden e-mail: alejandro.gonzalezvoyer[AT]ebc.uu.se Tel: ++46-18-471-2662 Web page: http://www.iee.uu.se/zooekol/default.php?type=personalpage&id=146&lang=en -----Original Message----- From: mesquitelist-bounces at mesquiteproject.org [mailto:mesquitelist-bounces at mesquiteproject.org] On Behalf Of Peter E. Midford Sent: den 24 april 2009 16:49 To: Chad Riopel Cc: mesquitelist at mesquiteproject.org Subject: Re: [Mesquitelist] Pagel94 vs Discrete rate parameter calculation Chad, I agree the results seem to disagree. I set up the Pagel94 module to be more aggressive (search longer) and cover a larger range of parameters than Discrete does typically. The likelihood surface for the search tends to be difficult and small differences in the search implementation may give large differences in the final result. As I recall from experimenting with Discrete while implementing the Pagel94 module, Discrete seemed to limit parameter values to be less than 50 (though that was only an inference on my part), so the solution Mesquite found would never be found by Discrete. You might try tweaking the likelihood iterations in Mesquite and see how it affects your answer. The other question is how do the reported likelihoods compare? If we could bound the range of parameter searches, would people find it useful? Hope this helps a bit, Peter On Apr 24, 2009, at 9:05, Chad Riopel wrote: > Hello, > I'm sorry to bother you all again but I'm confused as to > differences in results obtained between the Mesquite Pagel94 module > and results obtained by Discrete for maximum likelihood analysis. > The parameters for the 8 parameter dependent model based on Discrete > are: > > q12 = 1.52188 > q13 = 0.01739 > q21 = 0.0005 > q24 = 0.98443 > q31 = 0.21346 > q34 = 0.00061 > q42 = 4.13408 > q43 = 1.59419 > > and for the same data set from Mesquite I get: > > q12 = 2.0976873829078578E-4 > q13 = 26.837141573633282 > q21 = 0.0016498309403183997 > q31 = 4.245508561704672 > q24 = 12.481781495270134 > q34 = 5.122817481113189E-6 > q42 = 77.10900256167346 > q43 = 1.3547783960645563 > > The one parameter that concerns me is q13. In Discrete it's one of > the lowest parameters but in Mesquite it's one of the highest. This > has huge implications for the biological interpretation of my data. > I realize the manual states Mesquite calculates parameters with a > different optimizer which explains the differences, however it's > less clear on how to handle this situation. If anybody has > experienced this issue and/or knows how to deal with it, I'd > appreciate any feedback you might have. > > Thanks for your time, > > Chad > _______________________________________________ > Mesquitelist mailing list > Mesquitelist at mesquiteproject.org > http://mesquiteproject.org/mailman/listinfo/mesquitelist -- Peter Midford Mesquite Developer _______________________________________________ Mesquitelist mailing list Mesquitelist at mesquiteproject.org http://mesquiteproject.org/mailman/listinfo/mesquitelist From Heather.F.Smith at asu.edu Tue Apr 28 10:45:34 2009 From: Heather.F.Smith at asu.edu (Heather F. Smith) Date: Tue, 28 Apr 2009 10:45:34 -0700 Subject: [Mesquitelist] Random trees and weighting schemes Message-ID: <5288303a0904281045s193fe6d0u6a7a78ad2fa59cd2@mail.gmail.com> Dear Mesquite users, I am new to the program, and have run into an issue. I'm trying to summarize state changes across 10,000 random trees. The complication is that I would like to also incorporate a character state weighting scheme into the interpretations. I've created a step matrix with my desired weighting scheme, but when I run the analysis summarizing state changes over the random trees, the weighting scheme doesn't seem to be taken into consideration. When I do a test drastically altering the weights, such as weighting states 99:1, there is essentially no difference in the output. Does anyone have any advice about this issue? Has anyone else tried to run a similar analysis? Thanks very much! Sincerely, Heather Smith -- Heather F. Smith, Ph.D. Post-doctoral Associate Department of Basic Medical Sciences UA College of Medicine ? Phoenix 425 N. 5th Street Phoenix, Arizona 85004-2157 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mesquiteproject.org/pipermail/mesquitelist/attachments/20090428/4095ae6c/attachment.html From jeffrey.oliver at yale.edu Tue Apr 28 12:58:46 2009 From: jeffrey.oliver at yale.edu (Jeffrey C. Oliver) Date: Tue, 28 Apr 2009 15:58:46 -0400 Subject: [Mesquitelist] Random trees and weighting schemes In-Reply-To: <5288303a0904281045s193fe6d0u6a7a78ad2fa59cd2@mail.gmail.com> References: <5288303a0904281045s193fe6d0u6a7a78ad2fa59cd2@mail.gmail.com> Message-ID: Hi Heather, After setting up the Trace over trees, have you instructed Mesquite to use the step matrix for ancestral reconstructions? By default, I believe Mesquite will use unordered parsimony for reconstructions when the "Trace character over trees" is initialized. You can change the model from the menu by selecting Trace_Over_Trees > Source of parsimony models > Stored parsimony model. Mesquite will provide a list of available relevant models, which should include two default models (unordered and ordered) and your model. Select your model and the trace should be updated, using the values from your stepmatrix. Cheers, -Jeff 2009/4/28 Heather F. Smith > Dear Mesquite users, > > I am new to the program, and have run into an issue. I'm trying to > summarize state changes across 10,000 random trees. The complication is that > I would like to also incorporate a character state weighting scheme into the > interpretations. I've created a step matrix with my desired weighting > scheme, but when I run the analysis summarizing state changes over the > random trees, the weighting scheme doesn't seem to be taken into > consideration. When I do a test drastically altering the weights, such as > weighting states 99:1, there is essentially no difference in the output. > > Does anyone have any advice about this issue? Has anyone else tried to run > a similar analysis? Thanks very much! > > Sincerely, > Heather Smith > > -- > Heather F. Smith, Ph.D. > Post-doctoral Associate > Department of Basic Medical Sciences > UA College of Medicine ? Phoenix > 425 N. 5th Street > Phoenix, Arizona 85004-2157 > > _______________________________________________ > Mesquitelist mailing list > Mesquitelist at mesquiteproject.org > http://mesquiteproject.org/mailman/listinfo/mesquitelist > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mesquiteproject.org/pipermail/mesquitelist/attachments/20090428/cd7562b4/attachment.html From budd at embl-heidelberg.de Wed Apr 29 01:19:55 2009 From: budd at embl-heidelberg.de (Aidan Budd) Date: Wed, 29 Apr 2009 10:19:55 +0200 (CEST) Subject: [Mesquitelist] MESQUITE scripting - opening character and tree files Message-ID: Hi MESQUITEers I'm trying to write MESQUITE scripts that will load trees stored in around 100 different PHYLIP/NEWICK-format tree files, and binary character data (simple format) stored in the same number of external files (each of the 100+ trees is associated with one of the binary character datasets) such the appropriate tree/character-matrix pairs share the same taxa blocks The "trick" of doing the analysis via the MESQUITE GUI, and then looking at the MESQUITE scripting commands used to do the analysis doesn't work here - completely appropriately, MESQUITE simply writes appropriate taxa, tree, and charcter matrix NEXUS blocks based on the imported data, rather than specifying the script commands needed to reach this state. I'd greatly appreciate if someone could post an example script that would enable me to do something like this - I haven't been able to find an example to help me with this. Alternatively, perhaps this is just not feasible using MESQUITE, and I should rather put together a perl/python script that "simply" creates a NEXUS format file with appropriate tree/taxa/character-matrix blocks...? (Somehow it'd be a shame, though, if this were the way to do it - I'd rather avoid getting all the formating of these blocks right, parsing the input files etc.) Thanks Aidan -- ---------------------------------------------------------------------- Aidan Budd, PhD tel:+49 (0)6221 387 8530 EMBL - European Molecular Biology Laboratory fax:+49 (0)6221 387 8517 Meyerhofstr. 1, 69117 Heidelberg, Germany http://www.embl-heidelberg.de/~budd/ http://www-db.embl.de/jss/EmblGroupsHD/per_1807.html From budd at embl-heidelberg.de Wed Apr 29 01:25:26 2009 From: budd at embl-heidelberg.de (Aidan Budd) Date: Wed, 29 Apr 2009 10:25:26 +0200 (CEST) Subject: [Mesquitelist] running MESQUITE from the command line - specifying a project file to open Message-ID: Hi again, MESQUITEers Rather than tacking this on the end of my previous mail, thought it might be better to give it a separate thread... I'd like to be able, on my OSX machine, to run/execute MESQUITE from the command line in a way that specifies, via the command-line invocation, a NEXUS file to open in MESQUITE. I can start MESQUITE no problem from the command line, simply invoking the Run_Mesquite_OSX.command script However, knowing almost exactly nothing about java, I haven't been able to work out how to alter the java invocation to specify a MESQUITE NEXUS-format file to open on starting MESQUITE so that I can do something like: Run_Mesquite_OSX.command myMesquiteNexusProject.nxs so that MESQUITE will start, and then open "myMesquiteNexusProject.nxs" Any suggestions on how to do this would be much appreciated - or, of course, the advice to stop trying to do it as it's not appropriate/doesn't make sense! Thanks Aidan -- ---------------------------------------------------------------------- Aidan Budd, PhD tel:+49 (0)6221 387 8530 EMBL - European Molecular Biology Laboratory fax:+49 (0)6221 387 8517 Meyerhofstr. 1, 69117 Heidelberg, Germany http://www.embl-heidelberg.de/~budd/ http://www-db.embl.de/jss/EmblGroupsHD/per_1807.html