Merge pull request #686 from ARTbio/print_unique_counts

repenrich2 prints unique counts

tools/repenrich2/RepEnrich2.py

@@ -115,6 +115,12 @@ def run_bowtie(args):
     sumofrepeatreads += int(line[4])
 print(f"Identified {sumofrepeatreads} unique reads that mapped to repeats.")
 
+# print unique mapper counts
+with open("unique_mapper_counts.tsv", 'w') as fout:
+    fout.write("#element\tcount\n")
+    for count in sorted(counts):
+        fout.write(f"{count}\t{counts[count]}\n")
+
 # multimapper parsing
 if not paired_end:
     args_list = [(metagenome, fastqfile_1) for

tools/repenrich2/macros.xml

@@ -1,6 +1,6 @@
 <macros>
     <token name="@TOOL_VERSION@">2.31.1</token>
-    <token name="@VERSION_SUFFIX@">4</token>
+    <token name="@VERSION_SUFFIX@">5</token>
     <token name="@PROFILE@">23.0</token>
 
     <xml name="repenrich_requirements">

tools/repenrich2/repenrich2.xml

@@ -102,6 +102,7 @@
     </inputs>
 
     <outputs>
+        <data format="tabular" name="unique_mapper_counts" label="RepEnrich on ${on_string}: unique mapper counts" from_work_dir="unique_mapper_counts.tsv" />
         <data format="tabular" name="class_fraction_counts" label="RepEnrich on ${on_string}: class fraction counts" from_work_dir="class_fraction_counts.tsv" />
         <data format="tabular" name="family_fraction_counts" label="RepEnrich on ${on_string}: family fraction counts" from_work_dir="family_fraction_counts.tsv" />
         <data format="tabular" name="fraction_counts" label="RepEnrich on ${on_string}: fraction counts" from_work_dir="fraction_counts.tsv" />
@@ -114,6 +115,7 @@
             <param name="genomeSource" value="history"/>
             <param name="genome" value="chrY-1-500k.fa" ftype="fasta"/>
             <param name="repeatmasker" value="chrY-1-500k.fa.out" ftype="txt"/>
+            <output name="unique_mapper_counts" file="chrY_single_unique_mapper_counts.tab" ftype="tabular"/>
             <output name="class_fraction_counts" file="chrY_single_class_fraction_counts.tab" ftype="tabular"/>
             <output name="family_fraction_counts" file="chrY_single_family_fraction_counts.tab" ftype="tabular"/>
             <output name="fraction_counts" file="chrY_single_fraction_counts.tab" ftype="tabular"/>
@@ -125,6 +127,7 @@
             <param name="genomeSource" value="history"/>
             <param name="genome" value="chrY-1-500k.fa" ftype="fasta"/>
             <param name="repeatmasker" value="chrY-1-500k.fa.out" ftype="txt"/>
+            <output name="unique_mapper_counts" file="chrY_paired_unique_mapper_counts.tab" ftype="tabular"/>
             <output name="class_fraction_counts" file="chrY_paired_class_fraction_counts.tab" ftype="tabular"/>
             <output name="family_fraction_counts" file="chrY_paired_family_fraction_counts.tab" ftype="tabular"/>
             <output name="fraction_counts" file="chrY_paired_fraction_counts.tab" ftype="tabular"/>

tools/repenrich2/test-data/chrY_paired_unique_mapper_counts.tab

@@ -0,0 +1,146 @@
+#element	count
+A-rich	0
+ACCORD2_I-int	0
+ACCORD2_LTR	0
+ACCORD_I-int	0
+BARI1	0
+BATUMI_LTR	0
+BS	0
+BS2	58
+BURDOCK_I-int	0
+Baggins1	0
+Bica_I-int	38
+Bica_LTR	0
+CIRCE	0
+Chouto_I-int	0
+Copia1-I_DM	0
+Copia_I-int	25181
+Copia_LTR	647
+DIVER2_I-int	0
+DIVER2_LTR	0
+DM1731_I-int	6
+DM1731_LTR	0
+DM176_I-int	0
+DM412	0
+DM412B_LTR	0
+DMCR1A	4
+DMLTR5	0
+DMRT1A	0
+DMRT1B	86
+DMRT1C	0
+DNAREP1_DM	0
+DOC2_DM	0
+DOC3_DM	0
+FB4_DM	38
+FROGGER_I-int	0
+FW2_DM	0
+G3_DM	0
+G5A_DM	0
+G5_DM	0
+G6_DM	0
+GA-rich	0
+GTWIN_I-int	0
+G_DM	0
+Gypsy11_I-int	0
+Gypsy11_LTR	0
+Gypsy12_LTR	0
+Gypsy2-I_DM	2
+Gypsy2-LTR_DM	0
+Gypsy3_LTR	0
+Gypsy4_I-int	0
+Gypsy5_I-int	0
+Gypsy6A_LTR	0
+Gypsy6_I-int	26
+Gypsy8_I-int	0
+Gypsy8_LTR	0
+Gypsy9_I-int	0
+Gypsy_I-int	32
+Gypsy_LTR	1
+HELENA_RT	0
+HETA	24
+HMSBEAGLE_I-int	2
+IDEFIX_I-int	4
+IDEFIX_LTR	0
+Invader1_I-int	0
+Invader1_LTR	0
+Invader2_I-int	0
+Invader4_I-int	0
+Invader4_LTR	0
+Invader5_I-int	0
+Invader5_LTR	0
+Invader6_I-int	0
+Invader6_LTR	0
+MAX_I-int	49
+MAX_LTR	2
+MDG1_I-int	0
+MDG1_LTR	0
+MDG3_I-int	152
+MDG3_LTR	0
+MICROPIA_I-int	0
+MICROPIA_LTR	0
+Mariner2_DM	0
+NINJA_I-int	0
+NOMAD_I-int	0
+PROTOP_A	32
+PROTOP_B	0
+QUASIMODO2-I_DM	42
+QUASIMODO2-LTR_DM	0
+QUASIMODO_I-int	10
+QUASIMODO_LTR	2
+R1_DM	0
+ROOA_I-int	0
+ROOA_LTR	0
+ROVER-I_DM	381
+ROVER-LTR_DM	2
+S2_DM	0
+STALKER4_I-int	77
+STALKER4_LTR	4
+S_DM	48
+Stalker2_I-int	80
+Stalker2_LTR	2
+TART-A	4
+TART_B1	19
+TC1-2_DM	0
+TC1_DM	0
+TLD2	0
+TRANSIB1	0
+TRANSIB2	30
+ZAM_I-int	0
+_AACACA_n	0
+_AAT_n	0
+_ACAATAG_n	0
+_ACC_n	0
+_AGAGAAG_n	0
+_AGAGA_n	0
+_ATAAT_n	0
+_ATATATT_n	0
+_ATATTAT_n	0
+_ATTTTT_n	0
+_ATT_n	0
+_AT_n	0
+_A_n	0
+_CATA_n	0
+_CTTTT_n	0
+_GAGAA_n	0
+_GCCTTT_n	0
+_TAATAT_n	0
+_TAATA_n	0
+_TATAAAA_n	0
+_TATAA_n	0
+_TATCATG_n	0
+_TA_n	0
+_TGTTG_n	0
+_TTATATA_n	0
+_TTATAT_n	0
+_TTATA_n	0
+_TTA_n	0
+_TTCTT_n	0
+_TTC_n	0
+_TTTAT_n	0
+_TTTA_n	0
+_TTTC_n	0
+_TTTGA_n	0
+_TTTTAG_n	0
+_TTTTCTT_n	0
+_TTTTC_n	0
+_T_n	0

tools/repenrich2/test-data/chrY_single_unique_mapper_counts.tab

@@ -0,0 +1,146 @@
+#element	count
+A-rich	0
+ACCORD2_I-int	0
+ACCORD2_LTR	0
+ACCORD_I-int	0
+BARI1	0
+BATUMI_LTR	0
+BS	0
+BS2	29
+BURDOCK_I-int	0
+Baggins1	0
+Bica_I-int	19
+Bica_LTR	0
+CIRCE	0
+Chouto_I-int	0
+Copia1-I_DM	0
+Copia_I-int	12652
+Copia_LTR	134
+DIVER2_I-int	1
+DIVER2_LTR	0
+DM1731_I-int	3
+DM1731_LTR	0
+DM176_I-int	0
+DM412	0
+DM412B_LTR	0
+DMCR1A	2
+DMLTR5	0
+DMRT1A	0
+DMRT1B	35
+DMRT1C	0
+DNAREP1_DM	0
+DOC2_DM	0
+DOC3_DM	0
+FB4_DM	15
+FROGGER_I-int	0
+FW2_DM	0
+G3_DM	0
+G5A_DM	0
+G5_DM	0
+G6_DM	0
+GA-rich	0
+GTWIN_I-int	0
+G_DM	0
+Gypsy11_I-int	0
+Gypsy11_LTR	0
+Gypsy12_LTR	0
+Gypsy2-I_DM	1
+Gypsy2-LTR_DM	0
+Gypsy3_LTR	0
+Gypsy4_I-int	0
+Gypsy5_I-int	0
+Gypsy6A_LTR	0
+Gypsy6_I-int	12
+Gypsy8_I-int	0
+Gypsy8_LTR	0
+Gypsy9_I-int	0
+Gypsy_I-int	15
+Gypsy_LTR	0
+HELENA_RT	0
+HETA	12
+HMSBEAGLE_I-int	1
+IDEFIX_I-int	0
+IDEFIX_LTR	0
+Invader1_I-int	0
+Invader1_LTR	0
+Invader2_I-int	0
+Invader4_I-int	0
+Invader4_LTR	0
+Invader5_I-int	0
+Invader5_LTR	0
+Invader6_I-int	0
+Invader6_LTR	0
+MAX_I-int	27
+MAX_LTR	1
+MDG1_I-int	0
+MDG1_LTR	0
+MDG3_I-int	70
+MDG3_LTR	0
+MICROPIA_I-int	0
+MICROPIA_LTR	0
+Mariner2_DM	0
+NINJA_I-int	0
+NOMAD_I-int	0
+PROTOP_A	18
+PROTOP_B	0
+QUASIMODO2-I_DM	19
+QUASIMODO2-LTR_DM	0
+QUASIMODO_I-int	3
+QUASIMODO_LTR	1
+R1_DM	0
+ROOA_I-int	0
+ROOA_LTR	0
+ROVER-I_DM	188
+ROVER-LTR_DM	1
+S2_DM	0
+STALKER4_I-int	28
+STALKER4_LTR	0
+S_DM	25
+Stalker2_I-int	32
+Stalker2_LTR	2
+TART-A	2
+TART_B1	10
+TC1-2_DM	0
+TC1_DM	0
+TLD2	0
+TRANSIB1	0
+TRANSIB2	12
+ZAM_I-int	0
+_AACACA_n	0
+_AAT_n	0
+_ACAATAG_n	0
+_ACC_n	0
+_AGAGAAG_n	0
+_AGAGA_n	0
+_ATAAT_n	0
+_ATATATT_n	0
+_ATATTAT_n	0
+_ATTTTT_n	0
+_ATT_n	0
+_AT_n	0
+_A_n	0
+_CATA_n	0
+_CTTTT_n	0
+_GAGAA_n	0
+_GCCTTT_n	0
+_TAATAT_n	0
+_TAATA_n	0
+_TATAAAA_n	0
+_TATAA_n	0
+_TATCATG_n	0
+_TA_n	0
+_TGTTG_n	0
+_TTATATA_n	0
+_TTATAT_n	0
+_TTATA_n	0
+_TTA_n	0
+_TTCTT_n	0
+_TTC_n	0
+_TTTAT_n	0
+_TTTA_n	0
+_TTTC_n	0
+_TTTGA_n	0
+_TTTTAG_n	0
+_TTTTCTT_n	0
+_TTTTC_n	0
+_T_n	0