It looks like organism 46a is one of the most "primitive" ones, i.e. the one closest related to the common ancestor of most other species on the lower part of this tree. For the top part, I would guess 8a. However, the bootstrap values are low for most nodes, so it is hard to tell. Anything below 50% is unreliable, so do not trust the tree topology too much with those nodes. Also, predictions like these are very dependent on the depth of sampling of all extant organisms - if you are still missing many organisms, it is hard to get a reliable answer.

I hope this helps a bit.

It looks like organism 46a is one of the most "primitive" ones, i.e. the one closest related to the common ancestor of most other species on the lower part of this tree. For the top part, I would guess 8a. However, the bootstrap values are low for most nodes, so it is hard to tell. Anything below 50% is unreliable, so do not trust the tree topology too much with those nodes. Also, predictions like these are very dependent on the depth of sampling of all extant organisms - if you are still missing many organisms, it is hard to get a reliable answer.

I hope this helps a bit.

Hi Varsha

First of all I would assume that your aligment of sequences is really good and there are not problematic regions in the alignment.....because if the alignment is not good the output of phylogenetic analysis would also be affeceted....so make sure MSA is really good......

I would suggest you to use some other program also such as 'Treeview' to open the output file obtained from Phylip where you can actually arrange the tree branches in a manner where it is easy to explain (although the data remains same)......

I also agree with Elisabeth above that the bootstrap values looks low....try to refine it

by including more sequences if possible

bye

thank u sir and mam for your answers. I think my MSA is fine. I will try Treeview but i want to know that the method i was using for branch length calculations was right or not, as mentioned above in my question.

Hi Varsha,

I agree with Elisabeth and Jose. As far as I understand you which to compute an evolutionary distance. i think you have to use then a distance algorithm (neighbor, fitch,etc.), there are plenty of them. But you will get reliable relevant results only if your outgroup is obviously the right one and your sampling is extensive.

Hi Varsha, if your parameters for branch length optimization are correct, yes is OK to sum the distances.

Just another thing, is your tree rooted using a more "primitive" (I hate the use of this word in phylogenetics) outgroup?

I would take into account the comments above:

1. check you MSA masking out low complexity regions

2. try to imporve the bootstrap, but contrary to what of my colegues said, including more sequences of the same length (i.e. the same gene) typically would not help with the boostrap value, you will have to try to increase you data as in your alignment length, (i.e. concatenating genes, but make sure you partitionate the data). increasing the number of leafs only helps if there is a poor representation of the taxa in your original tree. The bootstrap can also be imporved by the step 1 mentioned here. Try a better alignment algorithm, what are you using?

3. make sure that you picked a good outgroup for the question you are trying to solve!

Cheers

The method you propose to compute the distance is correct: summing up all the branch-lengths between the two given nodes.

If you have to do this on many trees I suggest using ETE, which can automatically compute distances between any two nodes.

Other points discussed in some of the answers are relative to how much will you trust the branch-length estimation. This of course depends on the quality of the alignment, tree, and method used. Contrary to one of the suggestions above, I think distance-based methods such as Neighbor Joining, which are based on pairwise distances, are not good estimators of evolutionary distances. I would go for a maximum likelihood or bayesian approach that accounts for rate heterogeneity, making a prior test to ascertain which model fits best the data would also be advisable.

Hi Varsha,

the method is appropriate (and incidentally, proml is a maximum likelihood based method, so no need to change that), however, the question is tricky.

The tree you have built assumes a molecular clock and as a consequence the tree is (roughly) ultrametric, that is the distance from every tip to the root is equal. That is why you get the same answer for every species.

My suggestions are :

1. Try to get very good MSA

2. Select a proper sequence alignment method (depending on your sequence of interest) .

3. try to imporve the bootstrap value before calculating the distances.

> 2nd point is very important

> Try get the below books -

1) Bioinformatics: Sequence and Genome Analysis, Second Edition - David Mount

2) Biological Sequence Analysis: Probabilistic Models of Proteins and Nucleic Acids - Richard Durbin et al

3) Algorithms on Strings, Trees and Sequences: Computer Science and Computational Biology - Dan Gusfield

Thank you so much for your valuable suggestions..i will definitely solve my problem now.

1. You could try analyzing the distances in your tree using TreeRate; it can reroot your tree by various optimizations, and it can calculate all distances you may be interested in (use option BranchLength).

2. If you have low supports, you may not want to rely on a single tree. Instead you may want to analyze many plausible trees generated by, for instance, MrBayes.

First of all, Phylip produces unrooted tree topologies. You can indicate which taxon is your outgroup though and assume that the node where the outgroup connects to the ingroup is your root. Using the promlk programme in Phylip implies the use of a molecular clock, which means that the rates of evolution all throughout the tree are kept equal, and thus distances from root to leaf should be equal. You are correct in that to determine any distance between two taxa you indeed have to add up all branch lengths connecting the two taxa.

All extant taxa in your tree evolved from a common ancestor, and by using the molecular clock approach you also assume that all taxa underwent an approximate equal amount of changes... so there is no primitive or advanced taxon. All you can conclude from the tree is relationships among taxa (whether taxon "A" is more closely related to taxon "B" than "C"). To view the tree properly, open it in TreeView and use the unrooted "star" view.