home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!auvm!HKUCC.BITNET!VLDNASEQ
- X-Envelope-to: stat-l@MCGILL1.BITNET
- X-VMS-To: IN%"stat-l@mcgill1"
- Message-ID: <01GO2UOV4ZHC003B4G@hkucc.hku.hk>
- Newsgroups: bit.listserv.stat-l
- Date: Thu, 27 Aug 1992 14:09:00 +0800
- Sender: "STATISTICAL CONSULTING" <STAT-L@MCGILL1.BITNET>
- From: VLDNASEQ@HKUCC.BITNET
- Subject: Probability solution ?
- Lines: 67
-
- Dear Stat-L,
-
- A few days ago, I posted a question of probability on this board. The
- question is on the probability of finding all subclones in a subclone
- library of eight DNA fragments. In these few days, I received a few
- answers. However, I do not think that they are right. Most of them are
- on the right track, I must admit. The reason that I say that the answers
- I received are mathematically wrong is that:
-
- 1) mathematics is an exact thing. You get a definite probability of a
- number of subclones you screen, assuming that in your calculation,
- no approximation was taken. Then, if the number of subclones you
- screened is less than eight, the probability of finding all eight
- subclones will be equal to zero. I mean equal. So substituting
- numbers from 1 to 7, the probability solution must be equal to zero.
- This provides a rigorous test on the validity of the final equation
- deduced. Non of the solution I received pass through this test.
- 2) As the number of samples approach infinitive, probability will
- equal to one. This provides another test for the equation.
-
- After thinking the approach of the suggestions I received, I finally decided
- to solve the problem myself, though my mathematics level is only grade 11.
- I am happy to share with you my finger counting approach to solve the
- problem.
-
- For a subclone library with 8 different subclones, assume that there are
- infinitive numbers of bacterial colonies available, and distribution of the
- 8 different subclones are even. Let the probability of finding subclone be
- P(x) where x is the number of kinds of subclones obtained after picking n
- colonies.
-
- ^ means to the power of
- (3C2) means the combination of two kinds in three kinds, equal to 3
- . means times
- / means divided by
-
- for n is an integer which is not zero
-
- P(1) = (1/8)^n pick one particular kind of subclone in n
- times of picking
- P(2) = (2/8)^n -(2C1).P(1) probability of picking any two kinds of
- subclone minus that of picking only one kind
- P(3) = (3/8)^n -(3C2).P(2) - (3C1).P(1) picking three kinds minus two
- kinds only minus one kind only
- .
- .
- .
- .
-
- P(7) = (7/8)^n-(7C6).P(6)-(7C5).P(5)-(7C4).P(4)-(7C3).P(3)-(7C2).P(2)-(7C1).P(1)
-
- P(8) = 1 - (8C7).P(7) - (8C6).P(6) - (8C5).P(5) - (8C4).P(4) - (8C3).P(3)
- - (8C2).P(2) - (8C1).P(1)
-
- After substitution an simplification of the equation
-
- P(8) = [8^n - 8.7^n + 28.6^n - 56.5^n + 70.4^n - 56.3^n + 28.2^n - 8] / 8^n
-
- Note that this is a sequence, and the equation can be extended to any
- number of subclones in the library.
-
- Now I substitute n = 1 to 7 , P(8) = 0
- For n approach infinitive P(8) = 1
- So I am quite satisfied with this equation
-
- Please comment
- Wai Ming
-