| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
ProbeSet queries previously weren't dealing with aliases correctly,
because it was doing a MATCH/AGAINST against the ProbeSet.alias field,
but that field usually contains a list of gene symbols separated by
semi-colons (so it wouldn't detect the alias unless there was only a
single alias.
To fix this, I added some LIKE conditions, searching for the possible
variations. This is a little awkward, because I needed to make sure to
avoid a situation where, for example, an alias like 'LPD-1' matches a
search for 'PD-1'. I don't think the way it currently works is
efficient, but I don't know of any good alternative without changing the
way we store aliases in the database.
|
|
|
|
|
|
|
|
Moved the Heatmap display area above the table selection options (since
it didn't make sense to have it below those)
Added a button for clearing the Heatmap area and hiding the Heatmap
options
Added some margins to the Heatmap options
|
|
Heatmap options are now shown after the user clicks the button
I also reordered the Heatmap and Partial Correlation buttons, since it's
awkward to get the Heatmap options to display well if its button is
second. We'll probably completely rework this later, but for now this
looks a little less ugly.
|
|
Moved Partial Correlations and Generate Heatmap buttons to be lined up
below the other buttons and hide the Heatmap orientation selection (JS
will be added to show it after Heatmap is run)
|
|
Later this button should just be included with the other function buttons.
There's no need for it to be in a separate form, since it should be able
to take the same trait list input as all the other collection page
functions (with just the action url for the form changing)
|
|
Trait data caching wasn't working correctly because it didn't account
for the samplelist, causing caching to work incorrect in any situation
where the target dataset's samplelist wasn't the same as that of the
trait being correlated against. Trait data is stored as a dictionary
where the keys are trait IDs and values are *lists* of sample values.
This means that the caching needs to account for the exact same set of
samples; otherwise you'll end up with samples being mismatched (since
"the third sample with a value" for one dataset's trait might not be the
same as "the third sample with a value" for another dataset's trait).
To fix this, I added the samplelist to the functions that generate and
fetch the hash file. This will require more cache files, though, so this
should probably be reexamined later to make the code work with only a
single cache file for each dataset.
|
|
These checks are only relevant for python pre-2.2.
* wqflask/utility/svg.py: Remove python pre-2.2 checks.
|
|
* README.md: Replace GitHub actions badge with Laminar CI badge.
|
|
redis needs to be started and managed as a separate service. We create
a flag NO_REDIS so that it can be so. If NO_REDIS is not defined, the
present behavior of starting redis remains. Thus current use cases are
not broken.
* bin/genenetwork2: Allow for not starting redis with the NO_REDIS
flag.
|
|
Use a context manager so that the VERSION file is automatically closed
on exiting the block.
* etc/default_settings.py: Close VERSION file after opening.
|
|
There wasn't any search text for cis/trans searches including both
chromosome and exclusion zone, so I added somee
|
|
', Geno' was being added to the FROM clause, which is wrong becausee
Geno is included in the query via a JOIN (and caused an error making
those searches not work)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
