2
$\begingroup$

I was wondering given DNA samples, how can we create a phylogenetic tree? I mean I don't even want to have actual DNA sequence. Can I generate small samples, like reads of a length 100 or 1000 of nucleotide chains and figure out how they are related.

Suppose we have two reads of DNA samples, for brevity let's have them really small like: AAAGGTCTTGA and ATAAAGTGTA. I can run the sequence alignment algorithm on the strings and I will get {"A", {"", "T"}, "AA", {"G", "A"}, "GT", {"C", "G"}, "T", {"TG",""}, "A"} So basically what alighns then where you have indels or substitutions. I can easily now write up a function that calculates how close are these two strings by giving positive weights for matches strong negative weights for indels, etc. Now I want to know if these two guys with really short DNA are related or no. What is the mathematical measure of relatedness? And after I do Multiple Sequence Alignments and get similar results how do I build a tree from it.

$\endgroup$

closed as too broad by kmm, AliceD, James, rg255, March Ho Aug 25 '16 at 13:55

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    $\begingroup$ There's a lot of information about creating phylogenetic trees readily available online; e.g. evolution.berkeley.edu/evolibrary/article/phylogenetics_07 - can you add more specifically what you're having problems with? $\endgroup$ – arboviral Aug 9 '16 at 10:14
  • $\begingroup$ I am looking for rigorous mathematical explanation of how to use multiple sequence alignment of DNA sequences to build a phylogenetic tree. I can find the algorithm anywhere, but exactly how you build a Dendrogram after receiving output from sequencing? $\endgroup$ – Vahagn Tumanyan Aug 9 '16 at 10:24
  • $\begingroup$ I'd recommend editing any additional details or clarification into the question. I still don't think this answers my question ("what you're having problems with") though - if you have the algorithm (and I'd include this in your question, or a source to it; you may be using the wrong one) then what specifically are you having problems with? If you apply your algorithm to some sample data, what happens? $\endgroup$ – arboviral Aug 9 '16 at 10:53
  • $\begingroup$ Sorry, just noticed you're a new user - welcome to Biology.SE. I'd recommend taking the tour (biology.stackexchange.com/tour). Questions have to be specific and show what research you've already done, otherwise they won't get answered and may be closed. $\endgroup$ – arboviral Aug 9 '16 at 10:56
  • 1
    $\begingroup$ This is where it started: link.springer.com/article/10.1007%2FBF01734359 presuming you have an aligned sequence to begin with. $\endgroup$ – kmm Aug 9 '16 at 14:02
2
$\begingroup$

There is no single mathematical measure of relatedness.

In the context of phylogeny, you need a model of evolution in order to be able to evaluate some relatedness between your strings of nucleotides. Then you would have various option for how to build a tree so that it reflects this relatedness. See for instance the wikipedia page about computational phylogenetics: https://en.wikipedia.org/wiki/Computational_phylogenetics.

Through what kind of historical process were your sequences generated ? If there is no historical process underlying the generation of your sequences, then it is not meaningful to build a phylogeny with them. You can still define distances between them and use clustering techniques to build a tree, but that tree will not be a phylogeny.

$\endgroup$
1
$\begingroup$

Now I want to know if these two guys with really short DNA are related or no. What is the mathematical measure of relatedness

First of all this line of thought is dangerous, and creates a lot of room for misrepresentation, which have far reaching consequences when dealing with real DNA samples. Failed marriages are just one example.

Never infer phylogeny from short DNA sequences. You can read up on DNA fingerprinting here, which is used to infer ancestry and used in criminal prosecutions. But I think you already know about this.

Coming to inferring phylogeny on DNA samples where you have real DNA samples. For short sequences (upto 10kb, after that it becomes very time intensive), what you want is a Multiple sequence alignment such as Clustal Omega.

But the other procedure is to align entire genomes. Towards this end, you can check out this article from the UCSC team on the creation of their phylogenetics track, with the multiz package, the how to is described here. This procedure is based on an existing evolutionary model as stated by @bli. Finally, these alignments were used to generate conservation scores for the template genome (hg18,hg19 etc. etc.) using phastcons. If you go here, you will find the various multiz alignments between different phylogenetic trees, with the pattern multiz*way. Similarly, the inferred conservation from the respective multiz alignments with the use of phastcons and phylop is provided with phastcons*way or phylop*way. The help page for phastcons is here and for Phylop is here

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.