Hi Bernat,
my advice may not be that useful, but it may be better than the
silence so far...
Regarding the ordering of objects in hclust, if you're willing to do a
bit of hacking, have a look at the stats::hclust function; you will
see that the ordering is computed by a call to Fortran function
hcass2. That function could be used or perhaps adapted (it uses an
additional step that you will probably need to skip) to give you the
ordering. If you're good at understanding Fortran code, you may even
be able to re-write it directly in R.
Peter
On Tue, Jul 30, 2019 at 12:39 AM Bernat Gel Moreno <bgel at igtp.cat> wrote:
Hi,
For one of our packages (CopyNumberPlots) we'll need to read 10X CNV
data in H5 format. I've read in everything I need except for the cell
clustering tree. It's in a format called Z format produced by SciPy
hierarchical clustering. The format itself is relatively easy to parse
and not so different from hclust return objects so it would be possible
to create a small function to translate the Z notation into an hclust
object, if needed, but I'll need to figure out the "order" vector, since
it's not present in Z.
- Is the Z to hclust function available in any other package? Or
something equivalent to that?
- If I end up transforming it by hand in a custom function, Is there
a function somewhere to compute the order vector in an hclust object?
Thanks
Bernat