SEQanswers

Go Back   SEQanswers > Bioinformatics > Bioinformatics



Similar Threads
Thread Thread Starter Forum Replies Last Post
Say Hi to everybody! albraylee Introductions 0 11-01-2011 05:49 PM
hello, i am a novice but interested in NGS! Lyn Hsiong Introductions 0 09-26-2011 12:42 AM
Hello everyone! Lasia Introductions 0 08-23-2011 02:22 AM
Questions about solexa quality score! baohua100 Bioinformatics 23 05-20-2009 11:36 PM
Does SOLiD cost > 1 million USD !? salturki SOLiD 7 06-26-2008 01:16 PM

Reply
 
Thread Tools
Old 12-14-2011, 05:11 PM   #1
Godevil
Member
 
Location: Japan

Join Date: Feb 2011
Posts: 22
Default 【10 USD reward】Probability question for my experiment! Please help me!

I'm doing DNA library screening, and I meet a difficult probability question. I'm looking for some help here in the bioinformatics part.
I mean nothing of bribe or disrespect. If you don't like the thread title, please neglect it, as I found I cannot change the title after posting.

______________ We can absolute the biological question as the following pure math question:

There are [x] balls, including [3] red balls and [x - 3] grey balls. Every 96 balls are located randomly in one bag, so there are [n = x/96] bags. Now, I want to divide those bags into [m] equal teams. What is the probability that no more than one red ball locates in one team? (x & n are constant, m is the only variable.)

(I expect the answer and the explanation. Thank you very much! Please help me~~~~~)

_________________The real case is as follows:
I have a DNA library of 161280 clones, which are located in 1680 96-well plates. The library contains ~3x coverage of our model organism's genome, so there will be ~3 positive clones for one single copy gene in this library. My work is to find the positive clones of some genes of interest.

At first, I should divide those 96-well plates into [m] teams, and mix all samples in each team in order to detect which team has the positive clone we want. Then, using 3 or higher dimension based method to detect which well has the positive clone we want from that team. Because higher dimension screening method causes more false positive results, the less probability that one team has more than one positive clone, the better!

Accordingly, the point is more teams means less false positive results, however, it also means more money, more time, et. al.

Now, I want to calculate the probability that no more than one positive clone locates on one team, after dividing those plates into [m] teams. Finally, I can decide how many teams [m] I should make, according to the probability result above as well as money, time, labor, and some other factors.

(We should perform the screening work using a company's robot before Christmas and pay a lot of money, so I really want to calculate the probability as soon as possible in order to confirm the team number [m])

Last edited by Godevil; 12-15-2011 at 07:24 PM. Reason: Add the real case!!!
Godevil is offline   Reply With Quote
Old 12-14-2011, 05:56 PM   #2
nilshomer
Nils Homer
 
nilshomer's Avatar
 
Location: Boston, MA, USA

Join Date: Nov 2008
Posts: 1,285
Default

This is probably not the best place to find answers about random probability theory homework.
nilshomer is offline   Reply With Quote
Old 12-14-2011, 06:07 PM   #3
adaptivegenome
Super Moderator
 
Location: US

Join Date: Nov 2009
Posts: 437
Default

Quote:
Originally Posted by nilshomer View Post
This is probably not the best place to find answers about random probability theory homework.
Probably also not a good place to bribe or beg people to do their homework... perhaps as this site gets bigger we might need moderators to approve new threads? Or at least better classify them?
adaptivegenome is offline   Reply With Quote
Old 12-14-2011, 06:38 PM   #4
nilshomer
Nils Homer
 
nilshomer's Avatar
 
Location: Boston, MA, USA

Join Date: Nov 2008
Posts: 1,285
Default

Quote:
Originally Posted by genericforms View Post
Probably also not a good place to bribe or beg people to do their homework... perhaps as this site gets bigger we might need moderators to approve new threads? Or at least better classify them?
Personally, I try to look at all new posts, which is around 100-200 daily. We would need more moderators for classification and approvals.
nilshomer is offline   Reply With Quote
Old 12-14-2011, 06:41 PM   #5
adaptivegenome
Super Moderator
 
Location: US

Join Date: Nov 2009
Posts: 437
Default

Quote:
Originally Posted by nilshomer View Post
Personally, I try to look at all new posts, which is around 100-200 daily. We would need more moderators for classification and approvals.
I would volunteer
adaptivegenome is offline   Reply With Quote
Old 12-14-2011, 08:58 PM   #6
ECO
--Site Admin--
 
Location: SF Bay Area, CA, USA

Join Date: Oct 2007
Posts: 1,358
Default

I would love more moderators! If you're serious (or anyone else is...) let me know!

This thread is pretty funny. I would've made up a wrong but very believable answer...if only to teach the OP a lesson. Apologies in advance to the OP if this really is for his "experiment"

It also suggests a new way that users here could make some money...I'll have to think about that one a bit...
ECO is offline   Reply With Quote
Old 12-14-2011, 09:14 PM   #7
adaptivegenome
Super Moderator
 
Location: US

Join Date: Nov 2009
Posts: 437
Default

ECO, I sent you a PM. I am serious!
adaptivegenome is offline   Reply With Quote
Old 12-14-2011, 09:25 PM   #8
ECO
--Site Admin--
 
Location: SF Bay Area, CA, USA

Join Date: Oct 2007
Posts: 1,358
Default

Quote:
Originally Posted by genericforms View Post
ECO, I sent you a PM. I am serious!
You have a response...in your newfound usertitle...
ECO is offline   Reply With Quote
Old 12-14-2011, 09:35 PM   #9
adaptivegenome
Super Moderator
 
Location: US

Join Date: Nov 2009
Posts: 437
Default

Quote:
Originally Posted by ECO View Post
You have a response...in your newfound usertitle...
Thanks, ECO!
adaptivegenome is offline   Reply With Quote
Old 12-14-2011, 09:40 PM   #10
ECO
--Site Admin--
 
Location: SF Bay Area, CA, USA

Join Date: Oct 2007
Posts: 1,358
Default

Quote:
Originally Posted by genericforms View Post
Thanks, ECO!
Thank YOU sir.

I like that you immediately changed your usertitle back to Senior Member. Stealth moderator...I like it.
ECO is offline   Reply With Quote
Old 12-14-2011, 11:04 PM   #11
Godevil
Member
 
Location: Japan

Join Date: Feb 2011
Posts: 22
Default

Quote:
Originally Posted by nilshomer View Post
This is probably not the best place to find answers about random probability theory homework.
I'm sorry that I must post it on a wrong place. I'm just looking for some help.

I'm not so good at math, especially the combination and arrangement.

And I should perform my experiment as soon as possible, so I just want to get some help from here.

[This question is the real case in my experiment, and the 96 balls in one bag actually means the clones in a 96-well plate]
Godevil is offline   Reply With Quote
Old 12-14-2011, 11:09 PM   #12
Godevil
Member
 
Location: Japan

Join Date: Feb 2011
Posts: 22
Default

Quote:
Originally Posted by genericforms View Post
Probably also not a good place to bribe or beg people to do their homework... perhaps as this site gets bigger we might need moderators to approve new threads? Or at least better classify them?
Hi, I'm sorry I didn't mean to bribe. I just want to get some help for my experiment, because this experiment should be done as soon as possible. I've asked some pepole, but haven't get any correct answer.
Godevil is offline   Reply With Quote
Old 12-14-2011, 11:16 PM   #13
Godevil
Member
 
Location: Japan

Join Date: Feb 2011
Posts: 22
Default

Quote:
Originally Posted by ECO View Post
I would love more moderators! If you're serious (or anyone else is...) let me know!

This thread is pretty funny. I would've made up a wrong but very believable answer...if only to teach the OP a lesson. Apologies in advance to the OP if this really is for his "experiment"

It also suggests a new way that users here could make some money...I'll have to think about that one a bit...
This question is the real case in my experiment, and the 96 balls in one bag actually means the clones in a 96-well plate.

I'm not so good at this part of math, and I really need some help, as the experiment should be done as soon as possible.

Iíll change the title as soon as possible, if gays here don't like the bonus part. I don't mean any disrespect. I really need some quick help!
Godevil is offline   Reply With Quote
Old 12-14-2011, 11:29 PM   #14
Godevil
Member
 
Location: Japan

Join Date: Feb 2011
Posts: 22
Default

Quote:
Originally Posted by genericforms View Post
Probably also not a good place to bribe or beg people to do their homework... perhaps as this site gets bigger we might need moderators to approve new threads? Or at least better classify them?
Classification is very important. I just find I'm doing biological experiment, and I met a difficult math question, so I'm looking for some help here in the bioinformatics part.
Godevil is offline   Reply With Quote
Old 12-15-2011, 01:20 AM   #15
Simon Anders
Senior Member
 
Location: Heidelberg, Germany

Join Date: Feb 2010
Posts: 994
Default

I am actually willing to believe that you really are planning an experiment, but your way of asking the question is still very inconsiderate.

You know, we are all here because we are passionate not just about math and computer science but also about biology. If you really are designing an experiment tell us what the experiment is about. Otherwise, it is no fun to think about it.

Besides, the point of a public forum is to allow other readers to browse it and learn from the discussion. You are certainly not the only one who wants to randomly allocate controls over wells of microtiter plates to avoid spatial artifacts or do single-cell studies with stochastic cell sorting (just making two random guesses what this is about). Others with the same problem can profit from the discussion only if they know what it is really about.

Finally, in my experience, in 3/4 of the cases where a biologist asked me a question in an abstract way I have to change my advice once I learn what it is really about because the questioner omitted important details due to the abstraction.
Simon Anders is offline   Reply With Quote
Old 12-15-2011, 05:27 AM   #16
Godevil
Member
 
Location: Japan

Join Date: Feb 2011
Posts: 22
Default

Quote:
Originally Posted by Simon Anders View Post
Finally, in my experience, in 3/4 of the cases where a biologist asked me a question in an abstract way I have to change my advice once I learn what it is really about because the questioner omitted important details due to the abstraction.
Thank you for your advice, I should really tell the real case but not the transferred math question.
Godevil is offline   Reply With Quote
Old 12-15-2011, 06:55 AM   #17
krobison
Senior Member
 
Location: Boston area

Join Date: Nov 2007
Posts: 747
Default

I must confess, my usual route for solving these is to simulate them with a quick program (learning this skill can be valuable!); I'm always nervous I'll choose the wrong probability model. The obvious field for the right model would be hypergeometric, Poisson & binomial.

The sort of obvious guess is that since you have 3X coverage of your genome, there should be only about 3 positive wells in your entire collection. If that logic holds, then you can superpool quite a bit.

I would be tempted to initially generate pools for each plate, then combine those to form 96 superpools. You expect 3ish hits in the superpools. One more round of PCR lets you identify which plates have positive wells, and then you can just screen the plates.
krobison is offline   Reply With Quote
Old 12-15-2011, 10:44 AM   #18
Dbarker06
Junior Member
 
Location: Kentucky

Join Date: Sep 2010
Posts: 3
Default

Why not just arrange the 1680 plates into a 40X42 array. Then make pools of all the samples in each row of 40 and each column of 42. PCR these 82 pools. The results give you the row and column position of each positive 96 well plate in your 40x42 array. Then test each clone in the positive plates. You won't miss any positives--if there happen to be two positives in the same plate you will find that. Assuming that there are three positives,you will do less than 400 PCR reactions to find the individual positive clones.
Dbarker06 is offline   Reply With Quote
Old 12-15-2011, 11:42 AM   #19
ECO
--Site Admin--
 
Location: SF Bay Area, CA, USA

Join Date: Oct 2007
Posts: 1,358
Default

For what it's worth, I'm really glad this wasn't homework and actual experimental biology.

In the future (and anyone reading this contemplating doing the same bounty for help...)...please be honest. As Simon said, the fun is in the biology...you're much more likely to get help being honest (with or without a bounty).
ECO is offline   Reply With Quote
Old 12-15-2011, 07:35 PM   #20
Godevil
Member
 
Location: Japan

Join Date: Feb 2011
Posts: 22
Default

Quote:
Originally Posted by krobison View Post
I would be tempted to initially generate pools for each plate, then combine those to form 96 superpools. You expect 3ish hits in the superpools. One more round of PCR lets you identify which plates have positive wells, and then you can just screen the plates.
Actually, my question is just the relationship between that probability and the superpool number.
In your advice, you've already give the "96" as the superpool number, but I need the reason.

Last edited by Godevil; 12-15-2011 at 07:53 PM.
Godevil is offline   Reply With Quote
Reply

Tags
experimental design, probability

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off




All times are GMT -8. The time now is 08:40 PM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2019, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO