Merge pull request #175 from AlexandrovLab/rn7

v1.2.25: Add rn7 and resolve custom install tsb_BED path issue

Dockerfile

@@ -5,7 +5,7 @@ FROM python:3.8-slim
 WORKDIR /usr/src/app
 
 # Install SigProfilerMatrixGenerator from PyPI
-RUN pip install SigProfilerMatrixGenerator==1.2.23
+RUN pip install SigProfilerMatrixGenerator==1.2.25
 
 # Create a non-root user named 'spm_user'
 RUN useradd -m -s /bin/bash spm_user

SigProfilerMatrixGenerator/controllers/cli_controller.py

@@ -43,7 +43,7 @@ def parse_arguments_install(args: List[str]) -> argparse.Namespace:
     parser = argparse.ArgumentParser(description="Install reference genome files.")
     parser.add_argument(
         "genome",
-        help="The reference genome to install. Supported genomes include {c_elegans, dog, ebv, GRCh37, GRCh38, mm9, mm10, mm39, rn6, yeast}.",
+        help="The reference genome to install. Supported genomes include {c_elegans, dog, ebv, GRCh37, GRCh38, mm9, mm10, mm39, rn6, rn7, yeast}.",
     )
     parser.add_argument(
         "-l",
@@ -75,7 +75,7 @@ def parse_arguments_matrix_generator(args: List[str]) -> argparse.Namespace:
     parser.add_argument("project", help="The name of the project.")
     parser.add_argument(
         "reference_genome",
-        help="The name of the reference genome. Supported values {c_elegans, dog, ebv, GRCh37, GRCh38, mm9, mm10, mm39, rn6, yeast}.",
+        help="The name of the reference genome. Supported values {c_elegans, dog, ebv, GRCh37, GRCh38, mm9, mm10, mm39, rn6, rn7, yeast}.",
     )
     parser.add_argument("path_to_input_files", help="The path to the input files.")
 

SigProfilerMatrixGenerator/install.py

@@ -380,9 +380,9 @@ def install_chromosomes_tsb(genomes, reference_dir: ref_install.ReferenceDir, cu
 def install_chromosomes_tsb_BED(
     genomes, reference_dir: ref_install.ReferenceDir, custom
 ):
-    ref_dir = str(reference_dir.path)
-    tsb_bed_genome_dir = os.path.join(ref_dir, "chromosomes", "tsb_BED", genome, "")
     for genome in genomes:
+        ref_dir = str(reference_dir.path)
+        tsb_bed_genome_dir = os.path.join(ref_dir, "chromosomes", "tsb_BED", genome, "")
         # this `if` statement is strange. Usually the "chromosomes" folder is in
         # ref_dir + "references", not directly in ref_dir, so I would expect
         # the first clause to always be True

SigProfilerMatrixGenerator/references/chromosomes/transcripts/rn7/MT_transcripts.txt

@@ -0,0 +1,36 @@
+ENSRNOG00000030478	ENSRNOT00000040038	MT	1	68	1025	AY172581.9	Mt_rRNA
+ENSRNOG00000029171	ENSRNOT00000045072	MT	1	1026	1093	AY172581.3	Mt_tRNA
+ENSRNOG00000043866	ENSRNOT00000069133	MT	1	1094	2664	AY172581.24	Mt_rRNA
+ENSRNOG00000032112	ENSRNOT00000049156	MT	1	2665	2739	AY172581.14	Mt_tRNA
+ENSRNOG00000030644	ENSRNOT00000047550	MT	1	2740	3694	Mt-nd1	protein_coding
+ENSRNOG00000029301	ENSRNOT00000051008	MT	1	3695	3763	AY172581.4	Mt_tRNA
+ENSRNOG00000033545	ENSRNOT00000049396	MT	-1	3761	3831	AY172581.21	Mt_tRNA
+ENSRNOG00000032274	ENSRNOT00000042644	MT	1	3835	3903	AY172581.15	Mt_tRNA
+ENSRNOG00000031033	ENSRNOT00000040993	MT	1	3904	4942	Mt-nd2	protein_coding
+ENSRNOG00000029677	ENSRNOT00000046073	MT	1	4943	5008	AY172581.6	Mt_tRNA
+ENSRNOG00000033932	ENSRNOT00000044320	MT	-1	5010	5078	AY172581.22	Mt_tRNA
+ENSRNOG00000032609	ENSRNOT00000048525	MT	-1	5081	5152	AY172581.18	Mt_tRNA
+ENSRNOG00000031333	ENSRNOT00000045310	MT	-1	5185	5252	AY172581.10	Mt_tRNA
+ENSRNOG00000029954	ENSRNOT00000051860	MT	-1	5256	5321	AY172581.7	Mt_tRNA
+ENSRNOG00000034234	ENSRNOT00000050156	MT	1	5323	6867	Mt-co1	protein_coding
+ENSRNOG00000032882	ENSRNOT00000040283	MT	-1	6865	6933	AY172581.19	Mt_tRNA
+ENSRNOG00000031685	ENSRNOT00000051129	MT	1	6937	7004	AY172581.12	Mt_tRNA
+ENSRNOG00000030371	ENSRNOT00000043693	MT	1	7006	7689	Mt-co2	protein_coding
+ENSRNOG00000029070	ENSRNOT00000041978	MT	1	7693	7756	AY172581.1	Mt_tRNA
+ENSRNOG00000033299	ENSRNOT00000046201	MT	1	7758	7961	Mt-atp8	protein_coding
+ENSRNOG00000031979	ENSRNOT00000046108	MT	1	7919	8599	Mt-atp6	protein_coding
+ENSRNOG00000030700	ENSRNOT00000049683	MT	1	8599	9382	Mt-co3	protein_coding
+ENSRNOG00000029389	ENSRNOT00000048050	MT	1	9383	9450	AY172581.5	Mt_tRNA
+ENSRNOG00000033615	ENSRNOT00000041241	MT	1	9451	9798	Mt-nd3	protein_coding
+ENSRNOG00000032320	ENSRNOT00000052103	MT	1	9800	9867	AY172581.16	Mt_tRNA
+ENSRNOG00000031053	ENSRNOT00000044582	MT	1	9870	10166	Mt-nd4l	protein_coding
+ENSRNOG00000029707	ENSRNOT00000042928	MT	1	10160	11537	Mt-nd4	protein_coding
+ENSRNOG00000033957	ENSRNOT00000047162	MT	1	11538	11605	AY172581.23	Mt_tRNA
+ENSRNOG00000032578	ENSRNOT00000045499	MT	1	11606	11665	AY172581.17	Mt_tRNA
+ENSRNOG00000031667	ENSRNOT00000050396	MT	1	11665	11735	AY172581.11	Mt_tRNA
+ENSRNOG00000029971	ENSRNOT00000048767	MT	1	11736	13565	Mt-nd5	protein_coding
+ENSRNOG00000029042	ENSRNOT00000051268	MT	-1	13543	14061	Mt-nd6	protein_coding
+ENSRNOG00000032997	ENSRNOT00000049806	MT	-1	14062	14130	AY172581.20	Mt_tRNA
+ENSRNOG00000031766	ENSRNOT00000042098	MT	1	14136	15278	Mt-cyb	protein_coding
+ENSRNOG00000030339	ENSRNOT00000040611	MT	1	15280	15346	AY172581.8	Mt_tRNA
+ENSRNOG00000029145	ENSRNOT00000044639	MT	-1	15348	15415	AY172581.2	Mt_tRNA