Option to consider features all-exonic.

This is used for spike-ins, whose gtf file does not contain "exon"
features.

libhts/__init__.py

 from subprocess import check_output
 
-__copyright__ = "Copyright (C) 2020-2021 Blaise Li"
+__copyright__ = "Copyright (C) 2020-2022 Blaise Li"
 __licence__ = "GNU GPLv3"
-__version__ = 0.6
+__version__ = 0.7
 
 BEDTOOLS_VERSION = check_output(
     ["bedtools", "--version"]).decode("utf-8").strip().split()[-1]

libhts/libhts.py

@@ -141,10 +141,20 @@ class Gene():
                     self.exons, OVERLAP).nodes()))
 
 
-def gtf_2_genes_exon_lengths(gtf_filename):
-    """Returns a pandas DataFrame where union exon lengths are associated to gene IDs."""
+def gtf_2_genes_exon_lengths(gtf_filename, direct_len=False):
+    """
+    Return a pandas DataFrame where union exon lengths are associated to gene IDs.
+
+    If *direct_len* is set to `True`, features in the gtf file are assumed to be
+    "all exonic" and their length is taken directly without further controls.
+    This can be used for instance when dealing with spike-ins.
+    """
     gtf_file = open(gtf_filename, "r")
     gtf = BedTool(gtf_file)
+    if direct_len:
+        return pd.DataFrame(pd.Series(
+            {feature.attrs["gene_id"] : len(feature) for feature in gtf.features()},
+            name=("union_exon_len")).rename_axis("gene"))
     genes = {}
     for feature in gtf.features():
         feat_type = feature[2]