mgnify-pipelines-toolkit 0.1.6__py3-none-any.whl → 0.1.8__py3-none-any.whl

mgnify_pipelines_toolkit/analysis/amplicon/make_asv_count_table.py

@@ -277,6 +277,8 @@ def main():
     taxa_df = taxa_df.fillna("0")
     taxa_df = order_df(taxa_df)
 
+    asv_list = taxa_df.ASV.to_list()
+
     amp_reads = [read.strip() for read in list(open(amp, "r"))]
     headers = [read.split(" ")[0][1:] for read in list(open(headers, "r"))]
     amp_region = ".".join(amp.split(".")[1:3])
@@ -288,7 +290,7 @@ def main():
         counter += 1
         line_fwd = line_fwd.strip()
 
-        if line_fwd == "0":
+        if line_fwd == "0" or f"seq_{line_fwd}" not in asv_list:
             continue
 
         if headers[counter] in amp_reads:

mgnify_pipelines_toolkit/analysis/amplicon/primer_val_classification.py

@@ -58,19 +58,20 @@ def parse_args():
     return input, fasta, sample
 
 
-def get_amp_region(beg, strand, model):
+def get_amp_region(beg, end, strand, model):
     prev_region = ""
 
     for region, region_coords in model.items():
 
         region_beg = region_coords[0]
         beg_diff = region_beg - beg
+        end_diff = region_beg - end
 
         if strand == STRAND_FWD:
-            if beg_diff > 0:
+            if beg_diff > 0 and end_diff > 0:
                 return region
         else:
-            if beg_diff > 0:
+            if beg_diff > 0 and end_diff > 0:
                 return prev_region
 
         prev_region = region
@@ -92,6 +93,7 @@ def main():
             primer_name = line_lst[0]
             rfam = line_lst[3]
             beg = float(line_lst[5])
+            end = float(line_lst[6])
 
             if rfam == "RF00177":
                 gene = "16S"
@@ -116,7 +118,7 @@ def main():
             elif "R" in primer_name:
                 strand = STRAND_REV
 
-            amp_region = get_amp_region(beg, strand, model)
+            amp_region = get_amp_region(beg, end, strand, model)
             primer_seq = str(fasta_dict[primer_name].seq)
 
             res_dict["Gene"].append(gene)

mgnify_pipelines_toolkit/analysis/shared/get_subunits.py

@@ -102,7 +102,7 @@ def main():
     ]
     open_files = {}
     for record in SeqIO.parse(args.input, "fasta"):
-        model = "-".join(record.id.split("/")[0].split("-")[1:])
+        model = "-".join(record.id.split("/")[0].split("-")[-1:])
         if model in coding_rna:
             filename = pattern_dict[model]
         else:

{mgnify_pipelines_toolkit-0.1.6.dist-info → mgnify_pipelines_toolkit-0.1.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: mgnify_pipelines_toolkit
-Version: 0.1.6
+Version: 0.1.8
 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-0.1.6.dist-info → mgnify_pipelines_toolkit-0.1.8.dist-info}/RECORD

@@ -6,15 +6,15 @@ mgnify_pipelines_toolkit/analysis/amplicon/assess_inflection_point_mcp.py,sha256
 mgnify_pipelines_toolkit/analysis/amplicon/assess_mcp_proportions.py,sha256=aNucaUnYejl2Not4YLMBSzyYWGYJvYwLPZcFE94TIDc,5355
 mgnify_pipelines_toolkit/analysis/amplicon/classify_var_regions.py,sha256=kIuE2wo3FaFZw2-HRGxstKz29FyGuhqVDRhf_vPZgsA,19921
 mgnify_pipelines_toolkit/analysis/amplicon/find_mcp_inflection_points.py,sha256=EnsIrPGigsy8jVnjYgSECihhuquSJTgCi-k6fhusKYM,3547
-mgnify_pipelines_toolkit/analysis/amplicon/make_asv_count_table.py,sha256=pvMDfq-KA9LrD359fvM1uXbXa2Mow5Fja-1iUoVdSEg,8676
+mgnify_pipelines_toolkit/analysis/amplicon/make_asv_count_table.py,sha256=ICFR8Ci_VofQFykasiSWwOwL_SH64PVcROoenw5jifE,8751
 mgnify_pipelines_toolkit/analysis/amplicon/mapseq_to_asv_table.py,sha256=9QI6o85T4JPFq4EdKmnYzI6sxPLJG6t9W0xKiu24aqw,5035
-mgnify_pipelines_toolkit/analysis/amplicon/primer_val_classification.py,sha256=Zyo2Fb8ZFi8a5vNhLb-XpboNGWd2qyfMRaaGrXuFIgw,3634
+mgnify_pipelines_toolkit/analysis/amplicon/primer_val_classification.py,sha256=d_Mco92RRUXSq5-5oFlXC0ZO83kbxwOREoCCyA2glDc,3751
 mgnify_pipelines_toolkit/analysis/amplicon/remove_ambiguous_reads.py,sha256=8vwH6PY-XwMZhaUo08tOwdFsoREfNumvvDawTb9Y98U,3168
 mgnify_pipelines_toolkit/analysis/amplicon/rev_comp_se_primers.py,sha256=19NgCYE12bEvRBVibhZtZywwRiMdiBUBJjzL4by3_qo,1717
 mgnify_pipelines_toolkit/analysis/amplicon/standard_primer_matching.py,sha256=RDPsaWKf0wIDwvCHXyRCh2zSJf3y9E7uOhHjaAeX8bY,11099
 mgnify_pipelines_toolkit/analysis/shared/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 mgnify_pipelines_toolkit/analysis/shared/fastq_suffix_header_check.py,sha256=H5ccd1e_e5dk8vhVOvHLK1lknYbRPbnqPjULCYnU0FQ,4021
-mgnify_pipelines_toolkit/analysis/shared/get_subunits.py,sha256=xT-66JaeVA-jVh6l7PkvKSmMjnU6RsHz7sTiCehG_wk,4771
+mgnify_pipelines_toolkit/analysis/shared/get_subunits.py,sha256=xl5HduWtGPWiI9yqsjQ3itIzwHSxF2ig5KgjLXmj9EE,4772
 mgnify_pipelines_toolkit/analysis/shared/get_subunits_coords.py,sha256=DTX7S1P_BkGPEeDkbmUn1YoB247hpdNIe5rdFdRYDdA,1929
 mgnify_pipelines_toolkit/analysis/shared/library_strategy_check.py,sha256=XV1vjkjIHhzouM1k5hu_51XK_mgC_EOOGDN3mx4LOvc,1991
 mgnify_pipelines_toolkit/analysis/shared/mapseq2biom.py,sha256=exzWyuK0YxDiVSu4WX2H7g-uT5Y00w_EmrFqSHjRObU,5554
@@ -26,9 +26,9 @@ mgnify_pipelines_toolkit/constants/var_region_coordinates.py,sha256=jbOB_bTnW2TR
 mgnify_pipelines_toolkit/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 mgnify_pipelines_toolkit/utils/fasta_to_delimited.py,sha256=GbNT7clHso21w_1PbPpWKVRd5bNs_MDbGXt8XVIGl2o,3991
 mgnify_pipelines_toolkit/utils/get_mpt_version.py,sha256=zsQ4TuR4vpqYa67MgIdopdscsS0DVJdy4enRe1nCjSs,793
-mgnify_pipelines_toolkit-0.1.6.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-mgnify_pipelines_toolkit-0.1.6.dist-info/METADATA,sha256=eG7O-AjUoEx1OpdksKj9cfadbRUdzJ0TxT6ZRo2ANQ4,5859
-mgnify_pipelines_toolkit-0.1.6.dist-info/WHEEL,sha256=UvcQYKBHoFqaQd6LKyqHw9fxEolWLQnlzP0h_LgJAfI,91
-mgnify_pipelines_toolkit-0.1.6.dist-info/entry_points.txt,sha256=tCZ7ijAgfIn47xXGxNtoZbHTDyUfHjUzjXg-NBRlj6g,1646
-mgnify_pipelines_toolkit-0.1.6.dist-info/top_level.txt,sha256=xA_wC7C01V3VwuDnqwRM2QYeJJ45WtvF6LVav4tYxuE,25
-mgnify_pipelines_toolkit-0.1.6.dist-info/RECORD,,
+mgnify_pipelines_toolkit-0.1.8.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+mgnify_pipelines_toolkit-0.1.8.dist-info/METADATA,sha256=Uf-ukd_8rWTprPyipZQXTy4ZKdpxNezdwwPNwtNFbRk,5859
+mgnify_pipelines_toolkit-0.1.8.dist-info/WHEEL,sha256=OVMc5UfuAQiSplgO0_WdW7vXVGAt9Hdd6qtN4HotdyA,91
+mgnify_pipelines_toolkit-0.1.8.dist-info/entry_points.txt,sha256=tCZ7ijAgfIn47xXGxNtoZbHTDyUfHjUzjXg-NBRlj6g,1646
+mgnify_pipelines_toolkit-0.1.8.dist-info/top_level.txt,sha256=xA_wC7C01V3VwuDnqwRM2QYeJJ45WtvF6LVav4tYxuE,25
+mgnify_pipelines_toolkit-0.1.8.dist-info/RECORD,,

{mgnify_pipelines_toolkit-0.1.6.dist-info → mgnify_pipelines_toolkit-0.1.8.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (74.0.0)
+Generator: setuptools (75.2.0)
 Root-Is-Purelib: true
 Tag: py3-none-any