probability distributions for sampling

[caparks2]

I've found the sampling functions to be very helpful, especially because diversity measures can be influenced by changes in repertoire sizes between subjects. I appreciate that this is dealt with nicely in Immunarch by using robust measures of diversity, but when the more classical parametric measures are asked for it's useful to have the sampling functions available. I noticed the probability distributions used for sampling in the repSample() function were changed from the negative binomial to the multinomial. Any comment on the switch?

[ivan-balashov]

Hello, caparks2! My name is Ivan, I am a developer of the Immunarch package;

repSample uses multinomial sampling only with 'resample' method, but there is no any switching of distributions; other('downsample' and 'sample') methods use base::sample function inside; and I cannot find any mention about negative binomial sampling in commits;

Feel free If you have more details or other questions;
Cheers, Ivan

[caparks2]

Thanks–I must have been mistaken that the negative binomial was used previously.