You can download the adequate version of AFLPMax from the download menu at the left. Windows users should check if their computer has 32 or 64 bits and download the corresponding version. Windows and MacOS X versions are distributed as a compressed folder that includes the jar executable and a subfolder called binaries that must not be modified. Binaries contain all the necessary files for the correct functioning of AFLPMax. Linux version is distributed as a compressed folder that includes the jar executable and a subfolder called sources including a makefile. From the sources folder, the user can compile and install all source codes needed to run AFLPMax just by typing "make" from the linux console. AFLPMax requires the Java 6.0 (or higher) Runtime Environment freely available at http://www.java.com. Whatever the operating system, you should be able to run the program just by double clicking on the AFLPMax.jar file AFLPMax is a program aimed at finding the optimal number of AFLP (Amplified Fragment Length Polymorphism, Vos et al. 1995) sampled bands needed to reconstruct an accurate and well-supported AFLP-based phylogeny. Given a reference tree obtained a priori by any phylogenetic inference method and any source of information, AFLPMax provides the optimal number of AFLP bands that would be needed to reconstruct that tree with accuracy and, therefore, it is a tool useful to optimize resources and work for the reconstruction.

 

In brief, AFLPMax is a combination of programs that, using a tree defined by the user, performs the following tasks:

 

 

 

AFLPMax needs a reference tree to work. The tree has to be in Newick format (the same as used by PHYLIP), it has to be unrooted and it has to be written in only one line, without spaces. Here are two examples:

 

Example 1:

((Tax1:0.0125,Tax2:0.0125):0.0125,Tax3:0.0125,Tax4:0.025);

 

Example 2:

(Sp_1:0.033,Sp_2:0.033,(Sp_3:0.033,Sp_4:0.033):0.066);

 

Taxa names cannot start with a number but any alphanumeric character is admitted in any other position of the name. The first and only step needed to run the program is to load a tree. You can paste it directly into the INPUT TREE tab or you can load it from the File menu, option Open, also with the shortcut Ctrl+O. Once you have loaded a tree the Run button is activated and you can run an analysis using the default options or you can change them in the settings button and menu.

 

The reference tree defined by the user (see Input file section) is used to obtain DNA sequences with the program Seg-Gen (Rambaut & Grassly, 1997). Seq-Gen evolves sequences along a specific tree under a given evolution model (chosen by the user).

 

A computer program (aflp_seqgen) written in C is then used to simulate the AFLP technique. Each sequence corresponding to each taxon is cut separately with enzymes EcoRI and MseI, which are the typical enzymes used in AFLP studies. The program searches for all the restriction sites and returns all fragments that would result from the digestion. The output is the list of fragments sorted by their length, as it would appear in a real experiment. The in silico AFLP profiles are generated using the 4096 possible combination of three selective nucleotides for each enzyme. Another program (aflp_phylogeny), also written in C, is used to combine the information of all individuals and to construct the 1/0 presence/absence matrices. The program selects randomly the different number of bands used to reconstruct the phylogenies, starting from 100 and increasing by 100 each time (each set of 100 bands would correspond to a different primer combination) until a maximum of 1000 bands. Those numbers are chosen because 100 bands per AFLP profile are recommended for experimental data sets.

 

AFLP-based phylogenies are estimated with programs from the PHYLIP package (Felsenstein, 2005), using two of the most widely used methods in AFLP data sets: minimum evolution (ME) and maximum parsimony (MP). For the distance-based method (ME), each binary file of presence/absence is converted into a distance matrix with the program restdist using the Nei and Li distance (1979). The distance matrix obtained is used as input for the program fitch to infer the phylogeny under the ME algorithm. For the character based method (MP) the presence/absence matrix is used directly as input for the program pars. The user is allowed to generate the desired number of bootstrap pseudo-samples from the original data. To do that the program seqboot from the PHYLIP package is used. The obtained matrices are used to reconstruct the phylogenies with the same programs as before. A consense tree for MP and other for ME are obtained with the program consense with the minimum cut-off value indicated by the user. The phylogenetic inference can also be done with DNA sequences, which means that sequences of 10000 nucleotides are simulated with Seq-Gen and used directly to infer the tree. In this case the program dnadist is used to compute the distance matrix needed to reconstruct the phylogeny under the ME method with the program fitch. To obtain the MP tree the program dnapars is used.

 

Each estimated tree is then compared with the reference tree using the program Ktreedist (Soria-Carrasco et al., 2007), which takes into account both topology and branch length information of a phylogenetic tree. This program computes a K-score that measures overall differences in the relative branch length and topology of two phylogenetic trees after scaling one of the trees to have a global divergence as similar as possible to the other tree. The program also computes the symmetric difference or Robinson-Foulds (R-F) distance (Robinson and Foulds, 1981), which only takes into account the topology of the phylogenetic trees.

 

AFLPMax generates a table for each analysis and inference method showing the mean and standard error values obtained for the K-score and the R-F distance.
The outputs are two tables for each analysis, one corresponding to the ME results and another corresponding to the MP ones. Those tables are displayed in the corresponding tabs of the interface but are also saved in your computer as html files in a folder under your election. The names of the files start with ME_ or MP_ indicating the phylogenetic method used, the next part of the name refers to the analysis done (BAND: AFLP bands, BOOT: AFLP bands with bootstrapping or SEQ: DNA sequences). The last part of the name, RES, is common to the three possible analyses and is just to indicate that the file contains the table of results. For example, a file called ME_BAND_RES.html would contains the results of the AFLP analysis reconstructing the phylogenies with the ME inference method. Below you can find some examples of the tables of results, depending on the analysis done. If you run AFLPMax selecting AFLP analysis the output will show the number of bands used to reconstruct the phylogeny on the first column and the mean and standard error obtained from the replicates for the K-score and the R-F distance on the following ones. Values of K-score and R-F distance equal to zero indicate perfect match between the reference tree and the AFLP-reconstructed tree.

 

MP_BAND_RES:

num band	mean K	se K	mean RF	se RF
100	0.005794	0.000518	1.400000	0.221108
200	0.005521	0.000611	1.300000	0.260342
300	0.005034	0.000561	0.400000	0.221108
400	0.004514	0.000492	0.400000	0.266667
500	0.004157	0.000365	0.200000	0.200000
600	0.004545	0.000411	0.100000	0.100000
700	0.004402	0.000379	0.000000	0.000000
800	0.004186	0.000247	0.000000	0.000000
900	0.004398	0.000294	0.000000	0.000000
1000	0.004269	0.000356	0.000000	0.000000

 

ME_BAND_RES:

num band	mean K	se K	mean RF	se RF
100	0.005059	0.000611	2.400000	0.498888
200	0.004461	0.000487	1.000000	0.447214
300	0.003799	0.000451	0.200000	0.200000
400	0.002553	0.000174	0.600000	0.305505
500	0.002424	0.000235	0.600000	0.305505
600	0.002713	0.000345	0.000000	0.000000
700	0.002361	0.000337	0.000000	0.000000
800	0.002071	0.000205	0.000000	0.000000
900	0.001942	0.000224	0.000000	0.000000
1000	0.002322	0.000320	0.200000	0.200000