fix duplication rate calculation

docs/changelog.md

@@ -13,6 +13,8 @@
 ### Bugs fixed:
   - Fix #252: `tl.spectral` does not raise an error when the input matrix is in
     compressed column format, whereas it should expect a compressed sparse row format.
+  - Fix a bug in `pp.import_data` which produces incorrect duplication rates when
+    the input data contains mitochondrial reads.
 
 ## Release 2.6.0 (released March 9, 2024)
 

snapatac2-core/src/preprocessing/count_data/import.rs

@@ -141,9 +141,9 @@ where
     let mut qc = FragmentSummary::new(mitochrondrial_dna);
     let mut values = Vec::new();
     fragments.into_iter().for_each(|f| {
-        qc.update(&f);
         let chrom = &f.chrom;
         if genome_index.contain_chrom(chrom) {
+            qc.update(&f);
             let start = f.start as i64;
             let end = f.end as i64;
             let size = end - start;

snapatac2-core/src/preprocessing/qc.rs

@@ -199,10 +199,10 @@ impl<'a> FragmentSummary<'a> {
     }
 
     pub(crate) fn update(&mut self, fragment: &Fragment) {
+        self.num_total_fragment += fragment.count as u64;
         if self.mitochondrial_dna.contains(fragment.chrom.as_str()) {
             self.num_mitochondrial += 1;
         } else {
-            self.num_total_fragment += fragment.count as u64;
             self.num_unique_fragment += 1;
         }
     }

snapatac2-python/python/snapatac2/preprocessing/_basic.py

@@ -181,6 +181,8 @@ def import_data(
       temporary files is proportional to the number of records in the fragment file.
       For large fragment files, it is recommended to set `tempdir` to a location with
       sufficient space in order to avoid running out of disk space.
+    - The QC metrics are computed only for reads that are included by the `whitelist`
+      or `chrom_sizes`.
 
     Warning
     -------