http://tutorialforbioinformaticsfordham2011.blogspot.com/2011/12/kaks-tutorial-for-mega50.html
Using this method, I want to calculate dN/dS value, my file contains a root sequence, some homologous sequence with site mutations, however I got dS=0, how can I explain, is that mean this site rarely come with mutations?
You may have to use an alternative equation to avoid that zero in the denominator. So instead of saying ka/ks > 1 (which in your case it would be), you can say ka-ks>0. I found one paper that referenced this alternative form, and put the link below. I suspect that any statistical analysis you run would be affected by this new form of the equation, considering it measures the absolute instead of the relative difference between the two values.
Regardless, you can say that your ka/ks ratio is greater than one by some arbitrarily large amount and this supports a hypothesis of positive selection.
Comput Syst Bioinformatics Conf. 2006:335-40. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3090040/
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3090040/
There are different frequencies of synonymous and non-synomymous sites, with non-synonymous sites representing around 80% of all positions in a standard coding sequence. So if you have very few differences between the sequences it would be far more likely to have more/only non-synoymous sites affected just by chance. Assuming that not all non-synonymous changes have a high impact on fitness this could be big of a part of the answer to your result. This is why dN/dS ratio analyses are most useful for orthologous sequences that have not too few (but also not too much) differences.
In general, if you have a couple of non-synonymous differences between very closely related orthologous sequences and no synonymous ones, this can carefully be interpreted as signs of positive selection. Just consider that you defenitely would have taken the sign for true if you had just one synonymous substitution in addition... But, as already mentioned, the bioinformatic detection of signs of positive selection is only the first step. However, I would think you are aware of that.
Best, Hanno
Thanks to all of you, very appreciate for your advice. I found a paper
http://www.ncbi.nlm.nih.gov/pubmed/19081788
which hold conclusion that the relationship between selection and dN/dS is not necessarily a simple monotonic function, and it may be impossible to infer the selection pressure from the dN/dS measurement.
how do you think so?
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools