How to get parameter estimation r and p for Negative Binomial distribution using R?

Question

[bioinfo89]

Hi All,

I want to fit read count data to negative binomial distribution and estimate parameters using MLE method to get r=number of failures and p=prob of failures. I tried using fitdist function from fitdistrplus package, but I am unable to understand the estimated parameter size and mu if they are the same as r and p.

Has anyone used this function? Could you explain what does the estimated parameter mean? It is quite confusing to understand. Or suggest another function in R to do the same?

Any help will be much appreciated. Thank you!

[ATpoint]

May I ask for what type of data this is? There are already plenty of tools that make use of the NB, maybe there is one you can use out of the box.

[bioinfo89]

I am using ChIAPET read count data.

[ATpoint]

I would first look into something like https://bioconductor.org/packages/release/bioc/html/diffHic.html rather than developing custom approaches. Or the software it is based on, edgeR.

Answer 1

[Michael]

The R-packages fitdistr and fitdistrplus can do this.

[bioinfo89]

Yes, I tried this package, I am confused with the estimated parameters size specifically. I understood mu is the mean of failures (alternative parameterisation). But I am not sure what size mean. Is it number of successes or number of failures, equivalent to "r"?