onecodex library v1.0.0

This release contains breaking changes to plotting and classification results functions such as how

SampleCollection

is instantiated and how

metric

diversity_metric

, and

rank

are passed to

SampleCollection.to_df

and plotting functions.

Previously,

metric

and

normalize

were determined when a

SampleCollection

was instantiated. Now, those parameters are passed to

.to_df

and plotting functions separately, allowing users to change metrics at plot time.

Example:

##
# in 0.19.x
collection = onecodex.models.SampleCollection(samples, metric='readcount_w_children', normalize=True)
# produce a dataframe using readcount_w_children as the metric
collection.to_df()
# to plot a different metric, create a new SampleCollection
collection = onecodex.models.SampleCollection(samples, metric='abundance_w_children')
collection.plot_bargraph()

##
# in 1.0.0
collection = onecodex.models.SampleCollection(samples)
collection.to_df(metric='normalized_readcount_w_children')
# switch metrics without re-instantiating
collection.plot_bargraph(metric='abundance_w_children')
# for backwards compatibility, the following still works but prints a warning
collection = onecodex.models.SampleCollection(samples, metric='readcount_w_children')
# defaults to the metric provided when the collection was created
collection.to_df()

Added

metric
is now an argument for all plot and stats functions. By default, it is
auto
, which is determined by
SampleCollection.automatic_metric
based on existing logic
SampleCollection.__repr__
has been updated to differentiate it from lists:
<SampleCollection[Samples] length=60 : [<Samples 0b2d0b5397324841: "SRR4408293.fastq">, ...
Added type annotations to
onecodex.Api()
so that models are now known to static analyzers (e.g.,
ocx.Samples
)
SampleCollection.automatic_rank
now returns the best default
onecodex.enums.Rank
to use based on the contained results
SampleCollection.automatic_metric
now returns the best
onecodex.enums.Metric
value to use based on the contained results

Changed

You can change plotting metrics without re-instantiating the
SampleCollection
In alpha- and beta-diversity related functions (e.g.,
plot_mds
), the argument metric was changed to either
diversity_metric
or
distance_metric
. The
metric
argument in those functions now corresponds to the abundance metric (see
onecodex.enums.Metric
)
onecodex.notebooks.report
functions no longer raise a
OneCodexException
if used outside of IPython.

Removed

ClassificationDataframe.ocx
has been removed
SampleCollection()
no longer takes
metric
or
normalize
arguments. Those have been merged into a single
metric=
argument that takes a string or
onecodex.enums.Metric
enum value, which is now passed into dataframe and plotting functions directly
normalize
has been removed. Normalized metrics are now available as individual
Metric
enum values
Removed
test-python3-w-simple-json
from CI. It randomly started breaking. We don't think we need it anymore since we've moved away from Potion

Fixed

readcount
being used instead of the intended metric when clustering samples for PCA/Heatmap when plotting from a
SampleCollection
Plot title and table export column names have been updated to be more specific and match custom plots