help understanding methods involved in a phylogenetic analysis

[pluripotency]

I'm trying to read (with at least some comprehension) an article that is covering an analysis of the genomes (and partial genomes) of plants. The took about 150 different species and then attempted to reconstruct their evolution and how related they all are to one another. I need to form a presenation about the methods employed in their research, which is mostly analysis of the individuals' genes. This article is using a lot of words that I don't know and looking up the definitions of the words isn't proving that helpful to my comprehension either. A link to the article is here, with the passage titled "Phylogenetic Analysis" on page 9 being my focus.

 

I need an understanding of terms like Parsimony analysis, Partitioned Bremer support, and what exactly "(bootstrap = 100%)" means. I think this is mostly statistics related...?

 

Thanks so much, I know my questions are vague. Maybe we can just form a discussion on the subject?

Edited May 27, 2012 by totipotency

[CharonY]

Basically these are all terms that relate to statistical analyses of sequences. In very simple terms, if you compare sequences and see differences, you have to somehow come up with a means to quantify these changes in a meaningful way, as you assume that the smaller the distance, the closer the sequences are related to each other. There are a large number of statistical methods to do this that are based on different models. Parsimony analysis is one of these methods.

Further methods mentioned are then used to evaluate the quality and stability of this resulting tree.

[pluripotency]

Okay that makes sense, thank you.

 

Wouldn't you need to compare not only the raw sequences but also the significance of the genes affected? The sequence of an unimportant gene could have changed wildly because it's not selected for, thus telling the algorithm to classify it as another species when in reality it's really the same plant. Are one of the terms I mentioned supposed to account for that?

[CharonY]

It depends on the type of analyses you want to run. For phylogenetic analyses you would therefore look for genes that are under more or less universal selective pressure (that is why ribosomal sequences are useful).

[Arete]

Okay that makes sense, thank you.

 

Wouldn't you need to compare not only the raw sequences but also the significance of the genes affected? The sequence of an unimportant gene could have changed wildly because it's not selected for, thus telling the algorithm to classify it as another species when in reality it's really the same plant. Are one of the terms I mentioned supposed to account for that?

 

The assumption of most phylogenetic analyses attempting to infer the evolutionary relationships amongst organisms are that the genes being analyzed are evolving under a neutral model of selection. All of the available software will let you construct a phylogeny of a gene under selection, but making inferences from the subsequently produced tree might be erroneous.

 

You can test for the validity of this assumption using a test like Tajima's D, and you can also partition your data according to codon position and apply different substitution models to synonymous and non-synonymous substitution sites.

 

Of course, there's plenty of reasons one might wish to investigate the phylogenetic relationships between functional genes under selection, but the relationships between these genes and the overall genomes of the organisms they are from may well be quite different.

Edited May 29, 2012 by Arete

[pluripotency]

Thank you for that detailed response Arete.

 

 

In our paper, the team uses a product they've created called OrthologID to determine relationships between genomes. I assume in this case they've used orthologs in a weighting algorithm that classifies the organisms based on the differences in sequences that are largely the same among all genera. The orthologs the used are determined to be "functional"

 

 

CharonY that's an interesting point about the ribosomal DNA. In this paper, they're using nuclear DNA. In the past I know they've used mostly plastid DNA (ribosomal I guess in most cases?). Why is it that ribosomal DNA is more useful in creating an evolutionary tree. You said it's under a universal selective pressure, but I'm sure there are genes in nuclear DNA that could be identified and used as a more reliable means of determining evolutionary heritage. Because after all, doesn't nuclear DNA pass down the genome the ribosomes use?