Fix NA issues after Pandas update (#361)

* 🐛 fixes recursive_dbscan/pandas NA

* Fix hdbscan binning

* Update version

---------

Co-authored-by: Siddharth Uppal <suppal3@wisc.edu>

Author: chasemc

VERSION

@@ -1 +1 @@
-2.2.2
+2.2.3

autometa/binning/recursive_dbscan.py

@@ -186,7 +186,11 @@ def recursive_dbscan(
             coverage_stddev_cutoff=coverage_stddev_cutoff,
             gc_content_stddev_cutoff=gc_content_stddev_cutoff,
         )
-        median_completeness = filtered_df.completeness.median()
+        if filtered_df.empty:
+            median_completeness = float("-inf")
+        else:
+            median_completeness = filtered_df.completeness.median()
+
         if median_completeness >= best_median:
             best_median = median_completeness
             best_df = df
@@ -379,7 +383,11 @@ def recursive_hdbscan(
             coverage_stddev_cutoff=coverage_stddev_cutoff,
             gc_content_stddev_cutoff=gc_content_stddev_cutoff,
         )
-        median_completeness = filtered_df.completeness.median()
+        if filtered_df.empty:
+            median_completeness = float("-inf")
+        else:
+            median_completeness = filtered_df.completeness.median()
+            
         if median_completeness >= best_median:
             best_median = median_completeness
             best_df = df

autometa/binning/utilities.py

@@ -31,6 +31,8 @@
 
 import pandas as pd
 
+import numpy as np
+
 from typing import Iterable, Tuple
 
 from autometa.taxonomy.database import TaxonomyDatabase
@@ -183,7 +185,9 @@ def add_metrics(
         # redundant_marker_count = cluster_marker_counts.gt(1).sum(axis=1)
         # calculate completeness and purity and std. dev. metrics
         completeness = present_marker_count / reference_markers_count * 100
+        completeness = completeness.where(~np.isnan(completeness),pd.NA)
         purity = single_copy_marker_count / present_marker_count * 100
+        purity = purity.where(~np.isnan(purity), pd.NA)
         coverage_stddev = main_grouped_by_cluster.coverage.std()
         gc_content_stddev = main_grouped_by_cluster.gc_content.std()
         # merge metrics with given dataframe

nextflow.config

@@ -12,7 +12,7 @@ manifest {
     doi = "https://doi.org/10.1093/nar/gkz148"
     mainScript = "main.nf"
     nextflowVersion = ">=21.04.0"
-    version = "2.2.2"
+    version = "2.2.3"
 }