Add support for subsampling with replacement #33
Conversation
| } | ||
| } | ||
|
|
||
| void linked_sparse_transposed::transposed_subsample_with_replacement(const uint32_t n, const uint32_t random_seed) { |
There was a problem hiding this comment.
In looking close at this, I realized there is an added constraint when using "with replacement": it is necessary to filter out any sample whose counts are below the sampling depth. I realize this is not done by default in the biom.Table.subsample method but it arguably should be. The current work around is that we filter the table (i.e., biom.Table.filter) in advance of a with replacement call, and that is still viable to do here. What I recommend we do is get this merged first, then consider how to address filtering in a subsequent PR, does that sound good?
There was a problem hiding this comment.
Yes, this just implements subsampling with replacement as per biom.Table.
If you want a different semantics, we can change it to whatever is the most useful.
We do have the samples_count already, so it would be cheap.
There was a problem hiding this comment.
@wasade I think I have implemented what you requested.
No description provided.