I have undertaken a genetic diversity study on rice and some bootstrap values range as low as 10. I’m really looking to find out what you’d consider to be the most informative values
First you have to understand what is behind the bootstrap. It is a statistical procedure based on resampling your data several times (usually 100 to 1000), with random replacement of a very small fraction of the data. The end result is a majority-rule consensus tree showing the percentage a given group was recovered. As Niraj Singh pointed out, values below 50% should be ignored. Values above 50% should be taken into account comparatively: i.e. groups showing 95% are better supported than groups showing 90%, and so forth and so on. Therefore, there is no threshold, alpha, or "invalid" values, as Semir Gaouar mentioned above.
http://www.jstor.org/stable/2408678?seq=1#page_scan_tab_contents
You should consider bootstrap values >50 and delete lower values. You can discuss the confidence level of phylogenetic tree according to bootstrap values like Bootstrap (BS)
bootstrap values above 95% are regarded as invalid because it does not meet the minimum value of the alpha risk = 0.05
First you have to understand what is behind the bootstrap. It is a statistical procedure based on resampling your data several times (usually 100 to 1000), with random replacement of a very small fraction of the data. The end result is a majority-rule consensus tree showing the percentage a given group was recovered. As Niraj Singh pointed out, values below 50% should be ignored. Values above 50% should be taken into account comparatively: i.e. groups showing 95% are better supported than groups showing 90%, and so forth and so on. Therefore, there is no threshold, alpha, or "invalid" values, as Semir Gaouar mentioned above.
http://www.jstor.org/stable/2408678?seq=1#page_scan_tab_contents
Depends on the type of algorithm used for bootstrapping. For programs like RaxML, BS values of 70% or more are usually considered highly supported. There is a newer program called IQ-TREE that uses a different algorithm called Ultra-fast Bootstrapping and BS values derived from that is considered highly supported if it is 95% or more. Semir is wrong in stating that 95% values are invalid. In fact, it is the opposite. In most, if not all bootstrapping procedures, a 95% BS value is considered highly supported. These thresholds are not absolute but are a general rule of thumb that have been derived from simulation studies. Hope that helps.
Chan
Yes Dr Chan for ML using RaxML BS values should be >70% and for bayesian analysis >95% . There are several algorithm for phylogenetic analysis like Distance method NJ or UPGMA, MP, ML and Bayesian analysis. Among these 4 algorithm peoples are using 2 or 3 algorithm. I want to know that what criteria should be used for selection of algorithm like should I use NJ, ML, MP, Bayesian or all the algorithm?
that depends on what your objectives are. If you are looking to estimate a phylogeny, I would only either use ML or Bayesian inference.
- Badges
- Science topic
- Similar topics
- Biological Science
- Evolutionary Biology
- Phylogenetics