mgnify-pipelines-toolkit 0.1.3__py3-none-any.whl → 0.1.4__py3-none-any.whl

mgnify_pipelines_toolkit/analysis/amplicon/make_asv_count_table.py

@@ -223,6 +223,25 @@ def make_tax_assignment_dict_pr2(taxa_df, asv_dict):
 
     return tax_assignment_dict
 
+def generate_asv_count_dict(asv_dict):
+
+    res_dict = defaultdict(list)
+
+    for asv_id, count in asv_dict.items():
+
+        if count == 0:
+            continue
+
+        res_dict['asv'].append(asv_id)
+        res_dict['count'].append(count)
+
+    res_df = pd.DataFrame.from_dict(res_dict)
+    res_df = res_df.sort_values(by='asv', ascending=True)
+    res_df = res_df.sort_values(by='count', ascending=False)
+
+    return res_df
+
+
 def main():
     _TAXA, _FWD, _REV, _AMP, _HEADERS, _SAMPLE = parse_args()
 
@@ -250,23 +269,12 @@ def main():
     for line_fwd in fwd_fr:
         counter += 1
         line_fwd = line_fwd.strip()
-        fwd_asvs = line_fwd.split(",")
 
-        if paired_end:
-            line_rev = next(rev_fr).strip()
-            rev_asvs = line_rev.split(",")
-            asv_intersection = list(set(fwd_asvs).intersection(rev_asvs))
-
-            if len(asv_intersection) == 0:
-                continue
-
-            if len(asv_intersection) == 1 and asv_intersection[0] == "0":
-                continue
-        else:
-            asv_intersection = fwd_asvs
+        if line_fwd == '0':
+            continue
 
         if headers[counter] in amp_reads:
-            asv_dict[f"seq_{int(asv_intersection[0]) - 1}"] += 1
+            asv_dict[f"seq_{line_fwd}"] += 1
 
     fwd_fr.close()
     if paired_end:
@@ -285,6 +293,8 @@ def main():
         for tax_assignment, count in tax_assignment_dict.items():
             fw.write(f"{count}\t{tax_assignment}\n")
 
+    asv_count_df = generate_asv_count_dict(asv_dict)
+    asv_count_df.to_csv(f'./{_SAMPLE}_{amp_region}_asv_read_counts.tsv', sep='\t', index=False)
 
 if __name__ == "__main__":
     main()

{mgnify_pipelines_toolkit-0.1.3.dist-info → mgnify_pipelines_toolkit-0.1.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: mgnify_pipelines_toolkit
-Version: 0.1.3
+Version: 0.1.4
 Summary: Collection of scripts and tools for MGnify pipelines
 Author-email: MGnify team <metagenomics-help@ebi.ac.uk>
 License: Apache Software License 2.0

{mgnify_pipelines_toolkit-0.1.3.dist-info → mgnify_pipelines_toolkit-0.1.4.dist-info}/RECORD

@@ -6,7 +6,7 @@ mgnify_pipelines_toolkit/analysis/amplicon/assess_inflection_point_mcp.py,sha256
 mgnify_pipelines_toolkit/analysis/amplicon/assess_mcp_proportions.py,sha256=BCzLlfvRqiPC-YwzET901f_d0anYt1zpf5y0iOCQnvs,5191
 mgnify_pipelines_toolkit/analysis/amplicon/classify_var_regions.py,sha256=89LKH4rvqRydAEGvfaWqIClcitJ1Vbu7b5d4FApzGp4,18392
 mgnify_pipelines_toolkit/analysis/amplicon/find_mcp_inflection_points.py,sha256=5emeZjk8891IgyL5ygVcr8wMP-hGEJoEs2rcBbseWj0,3536
-mgnify_pipelines_toolkit/analysis/amplicon/make_asv_count_table.py,sha256=dZIygTbMZvVbSVBmFFAZz7x24oQEpvdEOTpTcnYAyoM,8444
+mgnify_pipelines_toolkit/analysis/amplicon/make_asv_count_table.py,sha256=-Kh9KlEuWWFAWUaNjkCxl-L3IdEMURLM3UKb2Tf81CM,8605
 mgnify_pipelines_toolkit/analysis/amplicon/mapseq_to_asv_table.py,sha256=lWVIhDxfsTuDzWjjUlMGx3RL7iD_Yy8m9Ppc9wjfCFg,4765
 mgnify_pipelines_toolkit/analysis/amplicon/primer_val_classification.py,sha256=F4ALSuH8N-0hHUqPCFwHgoAnteb2Ft3tUN9j6DaD5h8,3539
 mgnify_pipelines_toolkit/analysis/amplicon/remove_ambiguous_reads.py,sha256=yhDJFGD3t3TMfUlBCJGwzlK4IjFwm7Bporwp-aIM8uU,3139
@@ -20,9 +20,9 @@ mgnify_pipelines_toolkit/constants/regex_ambiguous_bases.py,sha256=oVeeCy33bY1WJ
 mgnify_pipelines_toolkit/constants/tax_ranks.py,sha256=fP97JvlUdxJBakJ694VresIY8-N3pcU99m7kZ9buKys,867
 mgnify_pipelines_toolkit/constants/thresholds.py,sha256=7J3caCikkEcLdKF4zSR0z8qMQw4-h9aSkSbFbS0LNg4,873
 mgnify_pipelines_toolkit/constants/var_region_coordinates.py,sha256=E8Cd3A1Hj9M95zw9Ut-2x8sE6_PlH6RJJEoikyZUMaQ,1303
-mgnify_pipelines_toolkit-0.1.3.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-mgnify_pipelines_toolkit-0.1.3.dist-info/METADATA,sha256=D8bYOR2kQZzJPdqtFkHj_Xd4axEHjzJPJXKAHtFj8L0,4950
-mgnify_pipelines_toolkit-0.1.3.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-mgnify_pipelines_toolkit-0.1.3.dist-info/entry_points.txt,sha256=K8nqnyAQG9jqHGgIfMIaCIe20u5a0FFCCqJWi4DoD2U,1306
-mgnify_pipelines_toolkit-0.1.3.dist-info/top_level.txt,sha256=xA_wC7C01V3VwuDnqwRM2QYeJJ45WtvF6LVav4tYxuE,25
-mgnify_pipelines_toolkit-0.1.3.dist-info/RECORD,,
+mgnify_pipelines_toolkit-0.1.4.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+mgnify_pipelines_toolkit-0.1.4.dist-info/METADATA,sha256=1s--DpsRsfAyM2HomRYTAfLF3shjuZ5o0CF-FoceZmw,4950
+mgnify_pipelines_toolkit-0.1.4.dist-info/WHEEL,sha256=Z4pYXqR_rTB7OWNDYFOm1qRk0RX6GFP2o8LgvP453Hk,91
+mgnify_pipelines_toolkit-0.1.4.dist-info/entry_points.txt,sha256=K8nqnyAQG9jqHGgIfMIaCIe20u5a0FFCCqJWi4DoD2U,1306
+mgnify_pipelines_toolkit-0.1.4.dist-info/top_level.txt,sha256=xA_wC7C01V3VwuDnqwRM2QYeJJ45WtvF6LVav4tYxuE,25
+mgnify_pipelines_toolkit-0.1.4.dist-info/RECORD,,

{mgnify_pipelines_toolkit-0.1.3.dist-info → mgnify_pipelines_toolkit-0.1.4.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: bdist_wheel (0.43.0)
+Generator: setuptools (70.3.0)
 Root-Is-Purelib: true
 Tag: py3-none-any