Manual - Christoph Heibl
Contents
1. LAGOPUS Divergence time estimation in R using the multidivtime approach Christoph Heibl and Natalie Cusimano Manual for the current version 1 4 4 as released on April 12 2010 Contents 1 Introduction 2 Required software 2 1 Getting R and the ape package onoo a a a a 2 2 Getting LAGOPUS 2 3 Getting phyloch optional e a 2 4 Getting PAML and multidivtime aooaa aa 0000 3 Preparing the data 3 1 DNA sequence alignments aoo aa a a 3 2 Phylogenetic trees 3 3 Age constraints 3 4 Creating the LAGOPUS input object o aa aa a a 4 Running LAGOPUS 4 1 Checking and editing the control files 2 4 2 Fixing the path to the executables 0 4 3 Starting the analysis 5 Acknowledgements oe KW WwWwwwnhn AAA A 1 Introduction Bayesian relaxed clock dating with PAML 4 1 Yang 2007 and multidivtime Thorne and Kishino 2002 is currently one of the most widely used molecular dating techniques However the procedure can be tedious and error prone because many different steps are involved The software is unintentionally user rude as put by Jeff Thorne in the multidivtime user manual In order to facilitate using PAML multidivtime for R users and others that might be convinced here we wrote a small R add on package that offers easy access to Bayesian relaxed clock dating once the input data are specified The package is called2. for upper bound which means a maximal age To sum up all this information we type node A lt c L t9 t3 72 node B lt ech S56 ta 12 my age constraints lt data frame rbind node A node B t12 mtii t10 Figure 1 An example tree with two internal nodes A and B which are represented in the fossil record 72 and 12 mya respectively Again you can check the data frame by typing my age constraints Note that age con straints which involve the outgroup taxa are not allowed in multidivtime This is be cause the outgroup will be pruned from the input tree in the MCMC analysis In this case LAGOPUS will issue a warning and terminate 3 4 Creating the LAGOPUS input object If you have created all these necessary objects i e sequence alignment phylogenetic tree and age constraints you will have to bundle them into an input object This is achieved by the mdt in function For an unpartitioned dataset this looks like x lt mdt in my sequences my tree my age constraints The function checks the consistency of the input objects and stores them in a list of class mdt in If you want to analyze a partitioned data set you will have to also specify a final tree This is necessary because the first step of parameter estimation using baseml requires that each tree match its corresponding alignment partition perfectly Subsequent steps estbranches and m
3. a file and load it prior to executation of LAGOPUS as described in Logical If transfer files TRUE all output files are trans ferred to an output folder which is created in the working direc tory of R If transfer files FALSE the output files will stay in the directory where the executables are A real number giving the mazimal difference between log Like lihood values of baseml and estbranches that will be accepted by LAGOPUS If the value of LogLCheck is exceeded the program assumes that one of the Likelihood estimations failed and will ter minate This mechanism is somewhat arbitrary and you should not blindly rely on it 5 Acknowledgements Jeff Thorne provided the algorithm for node assignment for phylogentic trees used by multidivtime Emmanuel Paradis reported several issues with multidivtime and gave advice for improvement Hanno Schafer offered his data set on Ctenoplectra bees for test ing and Susanne Renner gave useful comments on running multidivtime and tested the manual References Kishino H J L Thorne and W J Bruno 2001 Performance of a divergence time estimation method under a probabilistic model of rate evolution Mol Biol Evol 18 352 361 Thorne J L and H Kishino 2002 Divergence time and evolutionary rate estimation with multilocus data Syst Biol 51 689 702 Thorne J L H Kishino and I S Painter 1998 Estimating the rate of evolution of the rate of molecular evolution Mo
4. ape package Type read dna and read nexus data for details In ape DNA sequence alignments are stored as an object of mode matrix and class DNAbin Before proceeding you should check if the input file has been correctly parsed This can be done e g by typing summary my_sequences If the issued base frequencies seem reasonable usually everything is alright 3 2 Phylogenetic trees Now you proceed reading in your phylogenetic tree s You can use either my tree lt read tree my tree tre or my tree lt read nexus my tree nex both provided by the ape package The first command is for trees saved in the NEWICK format the latter for the NEXUS format The input tree can contain branch lenghts which are dropped automatically by LAGOPUS 3 3 Age constraints The age constraints are stored in a data frame object The example tree in figure 1 will demostrate the procedure Suppose you want to constrain nodes A and B to have been present at least 72 my and 12 my before present To identify both nodes you have to specify for each of them two terminal nodes that coalesce in these nodes The fossil record normally provides a minimal age for the node in question because we cannot be sure if there aren t any unknown older fossils out there and this characteristic is represented by L which stands for lower bound If you are completely sure that the node cannot be older e g strains of viruses etc you might also specify U
5. input object you are ready to execute multidivtime The interface is as follows multidivtime x file NULL start baseml part 1 runs 1 path NULL transfer files TRUE LogLCheck 100 plot TRUE Below follows a list with explanations of the arguments X file start part runs path transfer files LogLCheck An object of class mdt in containing DNA sequence alignment phylogeny and constraints see 3 A character string giving a name to the output files e g file mdt august2008 If left NULL the output files will be called LAGOUT The step to start with Can either be baseml estbranches or multidivtime If a step later than baseml is chosen the file argument must not be changed in order to allow LAGOPUS the accession of the previous output files If several partitions are to be analyzed LAGOPUS can be started or restarted on any partition The number of the partition is specified as an integer E g part 2 skips the first partition It is recommended to run Bayesian analyses several times to check for the convergence of the posterior probability distributions LAGOPUS offers a simple way to do this via the runs argument An integer gives the number of MCMC chains to run e g runs 3 A character string giving the path to the executables of basem1 paml2modinf estbranches and multidivtime which must all reside in the same directory see 2 4p It is convenient to store the path in
6. LAGOPUS in order to distinguish it in this manual from multidivtime LAGOPUS offers you e Input and control files are written automatically to the corresponding directories and without typos of course e Running baseml paml2modelinf estbranches multidivtime in a pipe e Age constraints are automatically assigned to the corresponding nodes Correct as signment of nodes can be verified in a cladogram that is printed to the screen with the age constraints mapped on it e The amount of evolution needed for calculating the mean and the standard devia tion of the prior distribution for the rate at the root node rtrate and rtratesd respectively is estimated by the median of branch lengths see the multidivtime read me e The output is summarized in an object of class mdt and can be visualized conveniently by the function plot mdt If this is what you are looking for read on 2 Required software 2 1 Getting R and the ape package Source code and binaries for the latest releases of R for all major platforms are distributed on http cran r project org We recommend to install the binary distribution Once installed you can launch the R console by doubleclicking on the R application in your applications folder or wherever you chose to install it or by typing open a R in your shell The ape package can be installed by the install packages command Make sure to install also the dependencies install packages ap
7. and installa tion of these programs please refer to the information provided in the respective manuals and read me s A helpful step for step manual for PAML multidivtime was written by Frank Rutschmann and can be downloaded at ftp statgen ncsu edu pub thorne bayesiandating1 5 pdf IMPORTANT In order to use LAGOPUS all the executables must be stored in one single folder The easiest way to achieve this is to copy the baseml executable into the multidis tribute folder On our computer this looks like this cp Applications paml14 src baseml Applications multidistribute 3 Preparing the data This section tells you how to prepare your data In order to start your analysis you need one or more alignments of DNA sequences one or more phylogentic trees with or without branch lengths and a table containing the age constraints to use on the data The following paragraphs show how to do this You can skip them if you know how to deal with phylogenetic data in R We suggest that you create a folder e g R files multidivtime where you drop your sequence and tree files 3 1 DNA sequence alignments First read in your alignment file s If you installed and loaded the phyloch package see above this can be done easily calling my sequences lt read align my sequences read align can parse files in NEXUS PHILIP and FASTA format Alternatively you can read in sequence data in the same formats by the functions provided by the
8. e dependencies TRUE For further information type install packages in the R console Upon successful instal lation ape is loaded by library ape 2 2 Getting LAGOPUS Download the gzipped LAGOPUS package from the LAGOPUS website http www christophheibl de mdt mdtinr htm1 It is not necessary to unpack the file just open a shell change to the directory where the gzipped file has been stored and type R CMD INSTALL LAGOPUS_1 4 2 tar gz Alternatively you can install LAGOPUS via the R console install packages LAGOPUS_1 4 2 tar gz repos NULL type source 2 3 Getting phyloch optional The phyloch package contains easy to use functions to import DNA sequences Other packages e g ape seqinr contain quite the same functionality and perhaps you prefere those If you decide to use phyloch download it from http www christophheib1l de Rpackages html and install it exactly like the LAGOPUS package 2 4 Getting PAML and multidivtime To carry out the Bayesian dating analysis baseml from Ziheng Yang s PAML package and paml2modelinf estbranches and multidivtime from the multidivtime package by Jeffrey Thorne and Hirohisa Kishino are required PAML is distributed on the website of Ziheng Yang http abacus gene ucl ac uk software paml html and multidivtime can be down loaded from Jeffrey Thorne s website nt tp statgen ncsu edu thorne multidivtime html For compilation
9. l Biol Evol 15 1647 1657 Yang Z 2007 Paml 4 a program package for phylogenetic analysis by maximum likeli hood Mol Biol Evol 24 1586 1591
10. ultidivtime relax this requirement so that you can select one tree for the final estimation of node ages and substitution rates mdt in recognizes the first tree in its argument list as final tree x lt mdt in seql seq2 seq3 final tr tri tr2 tr3 age constraints x lt mdt in seqi seq2 seq3 tri tr2 tr3 age constraints in the latter example trl is recognized as final tree 4 Running LAGOPUS 4 1 Checking and editing the control files Load the baseml and the multidivtime control files and edit them if necessary For example you should specify a mean rttm and standard deviation rttmsd for the prior distribution of the root node age i e the time interval from the root to present The choice of this value depends on the prior knowledge about the studied organisms data baseml ctr data multicntrl dat multicntrl dat rttm lt 84 set rttm to 84 mya multicntrl dat rttmsd lt 84 set rttmsd to 84 mya 4 2 Fixing the path to the executables The multidivtime function has one argument path whose default is NULL In order to save typing every time you start an analysis you can store the path in a RData object and load it prior to execution of multidivtime On CH s system this looks like this path lt Applications multidistribute save path file path RData before execution of multidivtime load Users R_files multidivtime path RDat 4 3 Starting the analysis Once you have created your
Download Pdf Manuals
Related Search