conifer-analysis 0.1.0

Post-process conifer output for downstream statistical analysis.

Post-process conifer output for downstream statistical analysis.

conifer-analysis uses dask in order to analyze conifer results in a distributed and out-of-memory fashion. This can be helpful when processing many such results.

Example

Say that you have a bunch of conifer results in a directory. You can generate a histogram of the confidence values per file (sample) and per taxa using the provided pipeline confidence_hist. Even when you work locally, it can be helpful to explicitly create a distributed client controlling the number of workers.

from dask.distributed import Client
from conifer_analysis import confidence_hist

client = Client(n_workers=8)

You can then visit the default dashboard in your browser to observe tasks live. Next, we run the pipeline which returns a pandas.DataFrame.

hist = confidence_hist("data/*.tsv")
hist.info()

As an example of the returned shape:

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7700 entries, 0 to 7699
Data columns (total 8 columns):
 #   Column       Non-Null Count  Dtype
---  ------       --------------  -----
 0   path         7700 non-null   category
 1   name         7700 non-null   category
 2   taxonomy_id  7700 non-null   category
 3   bin          7700 non-null   interval[float64, right]
 4   midpoints    7700 non-null   float64
 5   read1_hist   7700 non-null   int64
 6   read2_hist   7700 non-null   int64
 7   avg_hist     7700 non-null   int64
dtypes: category(3), float64(1), int64(3), interval(1)
memory usage: 385.3 KB

Install

It’s as simple as:

pip install conifer-analysis

If you want to observe tasks in the dask dashboard, you will need additional dependencies.

pip install conifer-analysis[dashboard]

Copyright

• Copyright © 2022, Moritz E. Beber.

• Free software distributed under the Apache Software License 2.0.