Dani Bodor
Dani Bodor
From [this blog post](https://blog.esciencecenter.nl/reduce-reuse-recycle-save-the-planet-one-github-action-at-a-time-4ab602255c3f). > # Reuse > > If you must run the CI, the best way to reduce its impact is limit the actual work that needs to...
deeprankcore's dependency on reduce is buggy. we will instead use an external package for adding hydrogens to the pdb files (in development). all code relying on reduce should be removed...
From [this comment](https://github.com/DeepRank/deeprank-core/pull/387#pullrequestreview-1372515081) (see below for examples): > We can either: > > Use forward slashes (/) or commas (,) with pathlib functions - both options work. The Path() object...
In #357 we decided to use distance cutoffs as a proxy for determining 1-3 and 1-4 pairing. A more precise way to determine this is using the adjacency matrix described...
We can consider adding an edge feature for polarity pairing (related to IRC node feature)
In pdb2sql v0.5.3 a new check was created that is being triggered by deeprankcore/tools/target.py when the `StructureSimilarity` object is created. I have temporarily suppressed it in PR #333. We should...
Would be nice if there can be setting for random seeds (for pytorch, numpy, random) to ensure consistency between experiments. Note that pdb2sql.transform.get_rot_axis_angle, also takes its own random
The buildgraph function (exclusively used in some unit tests) already existed before I worked on irc feature. Later I discovered that I largely recreated that function. The one used created...
Currently, each time we change something about the features, there is a potential that the tests break because the files in tests/hdf5/ are no longer representative. At this point we...
Currently, two modules from the neuralnets folder are not tested: - alignmentnet.py - ginet_nocluster.py Consider also resolving issue #259 while working on this issue.