GFOLD posterior distribution

Question

[tonja.r]

It seems I do not understand the posterior distribution described in this paper:

For a gene, given the observed read counts under two conditions, the posterior distribution of expression levels x1 and x2 under the two conditions can be calculated as above.

What values does lambda take in the formula of Post(lambda)? Does it take values from 0 to 1000, for example? And for each of this values a posterior probability is calculated and the most probable lambda is taken?

Answer 1

[jianxing.tongji]

Just consider lambda as the 'true' expression level and the read count from RNAseq as an observation. Therefore the value for lambda is ranging from 0 to infinity. For example, if a gene has expression 890 (=absolute molecular counts * a scalar constant), the observed read count is a random variable and could be 900,450,1300,... The only thing you know about the read count is that it follows Poisson distribution with parameter 890.