Hello!
I am looking to construct a library for cellular barcoding of hematopoietic stem cells. Are there probability based ratios I could use for determining the # possible barcodes I should have to get a library size that would be able to uniquely label X amount of HSC's?
For example Lets say the oligonucleotide I pick has 1,000,000 possibilities, how large should I make my library so that when I make a virus out of it, I am able to (or with a high probability) uniquely transfect, lets say, 10-20 cells.
This problem is actually a version of the coupon collector's problem i.e how much do I have to oversample to get X unique coupons?
http://mat.uab.cat/matmat/PDFv2014/v2014n02.pdf
The calculation will depend on what underlying assumptions you can make about the distribution e.g is there are equal probability of sampling the HSC? Is it a Poisson process or does it follow some other distribution? You could try and do a small sample and see what is barcoded to see what kind of distribution you are getting.
Yes, I would assume it is a poison process. I have run a couple of numbers given a variation of the coupon collector's problem, and with a barcode complexity set of 4,000 000, and I library size made from that of app. 100, the probability that picking from the library the same barcode is like 0.0001 percent. Then on top of it, transfecting it into HSC of a smaller size would (I imagine) further decrease the probability.
Though, in the process of creating the library, I take colonies on a plate (each having a unique barcoded plasmid), and I group them into one media and grow them together. Within the growing process, there may be some bias for certain barcodes to grow more. In addition, I will not have more then just 1 or each barcode, which may effect these probability.
I am probably going to make a library of 100 from a barcode complexity set of 4 000 000 and see what I get. Unless someone has done this before and could give me the numbers they used -- that would be helpful.
I find that papers describing the method are a little unclear on the library sizes they used .
- Badges
- Science method