bootstraping values is low , anyone familiar with DNAman program? - the bootstrap value is low, any way to increase it? (Oct/15/2006 )

hello all

i have 9 protein sequences of one particular gene from different strains of a same virus.
all the sequences are 99.6% similar ( DNA man program multiple alignment )
but when i creat the phylogenetic tree with bootstrap 10000 times,, the confidence level , bootstrap % is low

other genes of these strains show good bootstrap% values on the tree


does anyone know what that means? how can i change it?
i tried telling my proff that it only shows the confidnece level and the phylogeny is right
but he is interested in the bootstrap% being atleat around 90%


any suggestions???


any comment appreciated

regards and thanks

laxmi

---

If I remember correct the bootstrap values gives you the confidence of every branching of the tree, so the branching in your tree is not very good. Did you use a program that produces thousands of different tree possibilites to select the best (the consensus tree)?. This would be a possibility, I used Phylip for this some time ago. Other possibilites: Try other tree building methods (neighbor joining, parsimony, maximum likelihood) or improve your data, i.e. if there is not enough information to get suffcient separtion of the sequences or alignment of similar ones, you can for example use longer sequences.

---

Hi Laxmi,

There is no precise rule to say how high a bootstrap percentage has to be before you are sure that the group of species in question forms a "true" clade. However, values greater than 70% are often thought to be reasonably strong evidence. Bootstrap numbers need to be treated with caution. They are often a very useful indication of the reliability of different parts of a phylogenetic tree, but they DO NOT PROVE ANYTHING CONCLUSIVELY.

You could try optimisation criterions (neighbour joining, UPGMA or even maximum likelihood) for making judgements about trees but remember these are only predictions.

I think you should just explain the above statement to your professor.

Hope this helps

Good Luck!

---

Hi again,

Forgot to add this website....it's a phylogenetic tree drawing system called phylodraw.

http://pearl.cs.pusan.ac.kr/phylodraw/

It provides two clustering methods:

UPGMA (I wouldn't use this as the algorithm assumes a molecular clock)

NJ (neighbor joining - would recommend this)

Good Luck once again!

---

HI

hobglobin and sara


thanks for the suggestions , i will try that link and see too

i am using a program called DNAman, it uses NJ method for making the trees. and ya it calculates the bootstrap value by doing the sampling the no: of times one wants it to , in my case i done it 10000 times and got the value

i told my proff that its only a prediction and the value only shows how much the program can be sure about its prediction .
another explanation i came up is this :

the sequences are very similar 99.6 %
so i think the phylogenic tree was right , but the program couldnt say it for sure since there is so much little divergence ,, maybe thats why the bootstrap value is that low???

what do u guys think ? is it ok??

any comments appreciated

thanks and regards
laxmi

---

Hi Laxmi,

Yes you could say that. Low bootstrap values could also indicate lack of phylogenetic subdivisions. Depending on the distribution of variation among other taxa, the wrongly clustered short branch clade can give low bootstrap values overall.

That's all I can think of.

Hope this helps

Sara

---

hi thanks sara

i talked with proff and he accepetd it , thank god

thanks for the help and suggest

regards
laxmi

---

No probs Laxmi.....Glad it all worked out in the end

Sara

---