Handling zeros

[dawe]

Hello, I was applying propr-derived code on some scRNA-seq data and I got some results that prompted me to investigate the expression level of some interesting genes. Due to the fact some of the interesting genes in my settings are full of zeros, I wanted to check how zero are replaced in propr. According to the paper 0 is simply replaced with 1, this behaviour has been changed (I see there's a related issue #23 ) and 0 are replaced with the next smallest values. In the paper is

For simplicity, propr automatically replaces all zero values with 1 prior to log-ratio transformation, corresponding to the multiplicative replacement strategy

But reading the referenced paper (Martín-Fernández et al, 2003) the multiplicative replacement strategy requires the non-zero values to be replaced as well, multiplying them by the proportion of the non-zero values in the sample (eq. 6), to maintain the proportions.