sequencing prediction

[Tom Anderson]

On Sat, 2 Dec 2006, Jinchun Wang wrote:

> I have 20 different DNA duplexs in a mixture. After ligation,
> transformation, I got a plate of colonies. Now how can I predict how
> many colonies should be sequenced to find all 20 DNA?  If I already
> sequenced 100 colonies and got 12 DNA, is there some model to calculate
> how many more should be sequenced to find others?

If you look at it from the point of view of an individual insert (well,
species of insert), this should follow a binomial distribution:

http://en.wikipedia.org/wiki/Binomial_distribution

If you have 20 different inserts, and every colony has one insert, then
assuming all the inserts behave the same way, if you pick a colony at
random, there's a probability of 0.05 that it will have any given insert.
According to this calculator:

http://www.stat.tamu.edu/~west/applets/binomialdemo.html

After 100 colonies, that means there's a probability of 0.9941, ie
99.41%, that you will have found that insert.

That you are still missing eight inserts after having done this strongly
suggests that your inserts are not all behaving in the same way. I think
that means all bets are off in terms of statistics; if you had some idea
of how common those inserts would be, the binomial could tell you how many
colonies to sequence to find them, but since you haven't found them even
once yet, you don't have that information.

Do you know the sequences of the inserts? If so, is there a restriction
enzyme, or combination of enzymes, which will cut the twelve inserts you
have, but not the eight you don't? If so, make a mix of the constructs (by
repeating the ligation or growing up your whole population of
transformants), use the enzymes to destroy the ones you don't want, then
re-transform. The colonies from that transformation will hopefully be
enriched for the eight missing inserts.

tom

-- 
Tom Anderson, MRC Laboratory for Molecular Cell Biology, UCL, London WC1E 6BT
(t) +44 (20) 76797264   (f) +44 (20) 76797805   (e) thomas.anderson At ucl.ac.uk