Merge pull request #50 from YeoLab/11222024_pairedend_and_readspecific

11222024 pairedend and readspecific
YeoLab · Dec 3, 2024 · cf1eee9 · cf1eee9
2 parents 0c31f54 + 9c66360
commit cf1eee9
Show file tree

Hide file tree

Showing 9 changed files with 823 additions and 385 deletions.
diff --git a/marine.py b/marine.py
diff --git a/src/core.py b/src/core.py
@@ -149,7 +149,7 @@ def find_edits(bampath, contig, split_index, start, end, output_folder, barcode_
 
     output_file = '{}/{}_{}_{}_{}_edit_info.tsv'.format(edit_info_subfolder, contig, split_index, start, end)
     output_bedfile = '{}/{}_{}_{}_{}_edit_positions.bed'.format(edit_info_subfolder, contig, split_index, start, end)
-    
+
     remove_file_if_exists(output_file)
 
     with open(output_file, 'w') as f:        

diff --git a/src/utils.py b/src/utils.py
diff --git a/tests/integration_tests.ipynb b/tests/integration_tests.ipynb
@@ -2889,7 +2889,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 16,
+   "execution_count": 7,
    "id": "f5c3ed3e-13dd-4399-924e-3d1ac17ce387",
    "metadata": {},
    "outputs": [
@@ -3012,6 +3012,7 @@
       "\t >>> long_read_sc_test passed! <<<\n",
       "\n",
       "Checking that analyzing a single-cell dataset in 'bulk' mode (i.e. not specificying the 'CB' barcode) yields the exact same positions and base changes, but with counts and coverages aggregated rather than at a single-cell resolution\n",
+      "grouped_sc_rows: 62, bulk_rows: 62\n",
       "\n",
       "\t >>> single-cell and bulk on same dataset comparison passed! <<<\n",
       "\n",
@@ -3341,6 +3342,7 @@
     "bulk_folder = 'only_5_cells_bulk_mode_test'\n",
     "sc_5_cells = pd.read_csv('singlecell_tests/{}/final_filtered_site_info.tsv'.format(sc_folder), sep='\\t').sort_values(['position', 'strand_conversion'])\n",
     "bulk_5_cells = pd.read_csv('singlecell_tests/{}/final_filtered_site_info.tsv'.format(bulk_folder), sep='\\t').sort_values(['position', 'strand_conversion'])\n",
+    "\n",
     "print(\"Checking that analyzing a single-cell dataset in 'bulk' mode (i.e. not specificying the 'CB' barcode) yields the exact same positions and base changes, but with counts and coverages aggregated rather than at a single-cell resolution\")\n",
     "grouped_sc = pd.DataFrame(sc_5_cells.groupby(['contig', 'position', 'strand_conversion']).agg({'count': sum, 'strand_conversion': 'unique'}))\n",
     "grouped_sc.index.names = ['contig', 'position', 'c']\n",
@@ -3357,6 +3359,7 @@
     "    bulk_rows.append((r['contig'], r['position'], r['strand_conversion']))\n",
     "\n",
     "try:\n",
+    "    print(\"grouped_sc_rows: {}, bulk_rows: {}\".format(len(grouped_sc_rows), len(bulk_rows)))\n",
     "    assert(grouped_sc_rows == bulk_rows)\n",
     "    for bulk_item, grouped_sc_item in zip(bulk_rows, grouped_sc_rows):\n",
     "        assert(bulk_item == grouped_sc_item)\n",

diff --git a/tests/integration_tests_run.sh b/tests/integration_tests_run.sh
@@ -34,7 +34,7 @@ echo "SC tests scripts"
 ls -lh $MARINE/tests/$tests_folder/scripts/
 
 
-for t in "only_5_cells_test" "only_5_cells_bulk_mode_test" "long_read_sc_test" "edge_case_test" "edge_case_dist_filter_test"
+for t in "only_5_cells_test" "only_5_cells_bulk_mode_test" "only_5_cells_all_cells_coverage_test" "long_read_sc_test" "edge_case_test" "edge_case_dist_filter_test"
 
 do
     echo $t

diff --git a/tests/singlecell_tests/scripts/only_5_cells_bulk_mode_test.sh b/tests/singlecell_tests/scripts/only_5_cells_bulk_mode_test.sh
@@ -14,4 +14,4 @@ $MARINE/tests/singlecell_tests/only_5_cells_bulk_mode_test \
 --cores \
 4 \
 --strandedness 2 --verbose \
---contigs "9" --interval_length 32000000
+--contigs "9" --interval_length 32000000 --keep_intermediate_files
diff --git a/tests/singlecell_tests/scripts/only_5_cells_test.sh b/tests/singlecell_tests/scripts/only_5_cells_test.sh
@@ -15,4 +15,4 @@ $MARINE/tests/singlecell_tests/only_5_cells_test \
 4 \
 --barcode_tag "CB" \
 --strandedness 2 \
---contigs "9" --interval_length 32000000
+--keep_intermediate_files --contigs "9" --interval_length 32000000
diff --git a/tests/strandedness_tests/scripts/.ipynb_checkpoints/F1R2_pair_test-checkpoint.sh b/tests/strandedness_tests/scripts/.ipynb_checkpoints/F1R2_pair_test-checkpoint.sh
@@ -17,4 +17,4 @@ $MARINE/tests/strandedness_tests/F1R2_pair_test \
 --strandedness 2 \
 --contigs "chr17" \
 --sailor \
---num_intervals_per_contig 16
+--keep_intermediate_files
diff --git a/tests/strandedness_tests/scripts/F1R2_pair_test.sh b/tests/strandedness_tests/scripts/F1R2_pair_test.sh
@@ -17,4 +17,4 @@ $MARINE/tests/strandedness_tests/F1R2_pair_test \
 --strandedness 2 \
 --contigs "chr17" \
 --sailor \
---num_intervals_per_contig 16
+--keep_intermediate_files
-Original file line number
+Diff line change
@@ Expand Up / @@ -34,7 +34,7 @@ echo "SC tests scripts" @@
     ls -lh $MARINE/tests/$tests_folder/scripts/
-    for t in "only_5_cells_test" "only_5_cells_bulk_mode_test" "long_read_sc_test" "edge_case_test" "edge_case_dist_filter_test"
+    for t in "only_5_cells_test" "only_5_cells_bulk_mode_test" "only_5_cells_all_cells_coverage_test" "long_read_sc_test" "edge_case_test" "edge_case_dist_filter_test"
     do
         echo $t
@@ Expand Down @@