Multiple alignment feature #19
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…s not been exhaustively tested yet
Adds sorted alignments for HA and NA and a new (failing) test for identical PCA output regardless of the input order of sequences in multiple alignments.
Fixes a bug where PCA would produce incorrect output when input alignments were not sorted already by sorting alignments by sequence name before building the PCA matrix for each alignment.
Fixes a bug where we tried to calculate distances for PCA when these are not needed. This change fixes the preceding functional test that checks for an error when alignments don't have the same sequence names.
Makes the alignment input to pathogen-embed optional, but adds logic to require the alignment when running PCA or t-SNE with custom error messages for both cases. Adds functional tests for this behavior.
Sorts each distance matrix input alphabetically by sequence name (rows and columns) on load to ensure that the sum of all matrices correctly represents the same pairs of sequences. Adds an assertion to confirm that sequence names are identical for each distance matrix input. While we're improving this function, we also create an index of the sequence names for the final distance matrix data frame, so we don't have to set it later from alignment names. Adds a functional test to confirm that this behavior works as expected.
Adds/updates tests to ensure that distance matrices provided by the user or created on the fly from alignments are sorted and have the same samples. Refactors the logic for calculating distances from an alignment into its own function, so we don't have to copy the same code for pathogen-distance and pathogen-embed.
Adds a test and logic to error when distance matrix inputs are not square.
Adds tests for t-SNE behavior with multiple inputs when inputs are formatted correctly and incorrectly. Adds a check for matching record names in alignment and distance inputs to t-SNE when both are provided, to make the new failure mode test pass.
Explicitly sets the threading layer for Numba to a simple implementation that should always be available. UMAP uses Numba internally and, on some systems, the Numba threading layer can get set to an OpenMP (OMP) backend. Since Numba internally calls OMP with a function that is deprecated, OMP emits a deprecation warning. This warning is especially useless since we force UMAP to run with a single CPU. The main effect of this warning was to break our functional tests. For more details, see this Numba issue comment on GitHub: numba/numba#5275 (comment)
huddlej
force-pushed
the
multiple-alignment-feature
branch
from
051f8c1
to
9904998
Compare
Fixes a bug where the genetic distances plotted in the pairwise distance figure were not guaranteed to be integers. Adds figure outputs to a functional test which surfaced this issue originally.
Adds example commands to make an embedding from multiple inputs. This example comes after the quickstart which uses a single input.
Updates figures in the quickstart to match the new test data for HA and adds a figure for the HA/NA embedding clusters so readers can compare the results with the HA-only clusters.
|
@nandsra21 I finished adding tests, refining the code, and updating the documentation for this new feature. The README now includes an example tutorial for multiple input alignments. I'm going to test this out a bit with some other flu data, but I think this is nearly ready to merge and release. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.