Age | Commit message (Collapse) | Author |
|
Issue:
https://github.com/genenetwork/gn-gemtext-threads/blob/main/topics/gn1-migration-to-gn2/clustering.gmi
* While running tests for slink, to try and understand what it is doing in
order to write the appropriate tests for it, an issue arose that pointed a
blindspot in the former understanding of now `nearest` should work.
This commit fixes the issue found in both the expected data, and the code.
|
|
Issue:
https://github.com/genenetwork/gn-gemtext-threads/blob/main/topics/gn1-migration-to-gn2/clustering.gmi
* gn3/computations/slink.py: Add minimum code to pass new test
* tests/unit/computations/test_slink.py: new test
Add test to ensure that the new `slink` function return an empty list in
case and exception is raised.
Add the new `slink` function with minimum amount of code needed to pass the
test.
|
|
* gn3/computations/slink.py: add code to ensure new test passes
* tests/unit/computations/test_slink.py: new test
This one is a little weird: from
https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/heatmap/slink.py#L57-L63
It gets rid of the last coordinates in both the lists of the member
coordinates, and uses the remaining coordinates to find the shortest
members.
For example, given the following member coordinates:
- i=[0,1,2] and j=[5,7,9], it uses [0,1] and [5,7]
- i=[3,6,1] and j=[7,13], it uses [3,6] and [7]
to find the shortest distances.
I (fredmanglis) am not sure why it does it this way, since I'd have expected
it to use all the coordinates, however, since at this time we need to retain
bug-compatibility with the older code, I have done it as it is done in the
old code.
I also add a statement to raise an exception in the case where i and j are
not lists of integers, or integers
|
|
Issue:
https://github.com/genenetwork/gn-gemtext-threads/blob/main/topics/gn1-migration-to-gn2/clustering.gmi
* gn3/computations/slink.py: add code to pass new test
* tests/unit/computations/test_slink.py: new test
Given a list of members in a group, and a coordinate for a member in the
same group, find the distance of the closest member from the given
coordinate in the group.
|
|
* gn3/computations/slink.py: Add code to compute the distance given the
coordinate of both members on the parent list/tuple
* tests/unit/computations/test_slink.py:
* Change the name of the tests to more closely correspond to the business
requirement the test is checking for
* Update the comments to indicate some more things that might need to be
done in the future
|
|
Issue:
https://github.com/genenetwork/gn-gemtext-threads/blob/main/topics/gn1-migration-to-gn2/clustering.gmi
* gn3/computations/slink.py: check that all distances between the 'somethings'
are all either zero or positive.
* tests/unit/computations/test_slink.py:
* Remove data with all distances positive or zero, since it would fail the
test
* Change the expected message to more closely correspond to the business
logic
|
|
Issue:
https://github.com/genenetwork/gn-gemtext-threads/blob/main/topics/gn1-migration-to-gn2/clustering.gmi
* gn3/computations/slink.py: check that the distance from child A to B is the
same as distance from child B to A. If not, throw an exception.
* tests/unit/computations/test_slink.py:
* Change the name of the test to more closely correspond to the business
logic being tested.
* Update the data in a separate test such that it does not error out due to
failing to fulfill the expectations of separate requirement.
- pass tests
- Rename test
- Fix errors: distances same both directions
|
|
Issue:
https://github.com/genenetwork/gn-gemtext-threads/blob/main/topics/gn1-migration-to-gn2/clustering.gmi
* gn3/computations/slink.py: Check that a child's distance from itself is
zero. If not, throw an exception.
The children lists are a list of distances of "something" from other
"somethings". There is still some need to establish what those "somethings"
are, so that the test names can reflect the ideas that are actually being
tested for.
* tests/unit/computations/test_slink.py: Change the name of the test so that
it more closely corresponds to the business logic it is actually testing,
and not the mechanics of testing the idea.
|
|
Issue:
https://github.com/genenetwork/gn-gemtext-threads/blob/main/topics/gn1-migration-to-gn2/clustering.gmi
* gn3/computations/slink.py: Add dummy `nearest' function
* tests/unit/computations/test_slink.py: Add some tests
This commit adds tests to try and reproduce the working of the `nearest'
function in:
https://github.com/genenetwork/genenetwork1/blob/master/web/webqtl/heatmap/slink.py
This commit might not yet have extracted all the expected behaviour of the
`nearest' function, and therefore, there is a possibility of a later commit
augmenting the work in this commit.
|