I've just started using deseq2 and have noticed that some p adjusted values are appearing as NA in my output CSV. The issue with this is the package excludes any genes with an adjusted p-value = NA from the total ranking count and therefore generates an inaccurate adjusted p-value.

i.e. total genes measured = 520

genes with adjusted p-value = 0.000etc, n = 480

genes with adjusted p-value = NA, n = 40

Example where genes with adjusted P-value = NA are excluded: Most significantly altered gene, p-value = 0.0001, adj.p-value = 0.0001*(480/1)=0.048

Example where genes with adjusted P-value = NA is included: Most signigicantly altered gene, p-value = 0.0001,adj.p-value = 0.0001*(520/1)=0.052

Should we not still include all genes that had read counts in the total ranking for a correct BH adjustment?

This is discussed in the vignette and the DESeq2 paper. You can also read about IHW which is a more sophisticated approach.