Merge pull request #687 from ARTbio/fix_repenrich2_bammanip

Fix repenrich2 bam manipulation
ARTbio · Apr 22, 2024 · a9af128 · a9af128
2 parents b1761de + d15947d
commit a9af128
Show file tree

Hide file tree

Showing 9 changed files with 338 additions and 338 deletions.
diff --git a/tools/repenrich2/RepEnrich2.py b/tools/repenrich2/RepEnrich2.py
@@ -197,14 +197,14 @@ def run_bowtie(args):
 # print class-, family- and fraction-repeats counts to files
 with open("class_fraction_counts.tsv", 'w') as fout:
     for key in sorted(classfractionalcounts):
-        fout.write(f"{key}\t{classfractionalcounts[key]}\n")
+        fout.write(f"{key}\t{round(classfractionalcounts[key], 2)}\n")
 
 with open("family_fraction_counts.tsv", 'w') as fout:
     for key in sorted(familyfractionalcounts):
-        fout.write(f"{key}\t{familyfractionalcounts[key]}\n")
+        fout.write(f"{key}\t{round(familyfractionalcounts[key], 2)}\n")
 
 with open("fraction_counts.tsv", 'w') as fout:
     for key in sorted(fractionalcounts):
         fout.write(f"{key}\t{repeat_ref[key]['class']}\t"
                    f"{repeat_ref[key]['family']}\t"
-                   f"{fractionalcounts[key]}\n")
+                   f"{round(fractionalcounts[key], 2)}\n")
diff --git a/tools/repenrich2/macros.xml b/tools/repenrich2/macros.xml
@@ -1,6 +1,6 @@
 <macros>
     <token name="@TOOL_VERSION@">2.31.1</token>
-    <token name="@VERSION_SUFFIX@">5</token>
+    <token name="@VERSION_SUFFIX@">6</token>
     <token name="@PROFILE@">23.0</token>
 
     <xml name="repenrich_requirements">

diff --git a/tools/repenrich2/repenrich2.xml b/tools/repenrich2/repenrich2.xml
@@ -38,16 +38,16 @@
             bowtie2 -x $index_path -p \${GALAXY_SLOTS:-4} input.fastq
                 | samtools sort -@ "\${GALAXY_SLOTS:-4}" -T tmp -O bam -o aligned.bam 2>&1 &&
             samtools view -@ "\${GALAXY_SLOTS:-4}" -F 4 -b -q 38 aligned.bam -o unique.bam &&
-            samtools view -@ "\${GALAXY_SLOTS:-4}" -h -F 4 -b aligned.bam \
-                | samtools view -@ "\${GALAXY_SLOTS:-4}" -U -b -q 38 - \
-                | bedtools bamtofastq -i /dev/stdin -fq multimap.fastq &&
+            samtools view -@ "\${GALAXY_SLOTS:-4}" -F 4 -b aligned.bam
+                | samtools view -@ "\${GALAXY_SLOTS:-4}" -U filt_aligned.bam -bq 38 - > /dev/null &&
+            bedtools bamtofastq -i filt_aligned.bam -fq multimap.fastq &&
         #else:
             bowtie2 -x $index_path -p \${GALAXY_SLOTS:-4} -1 input.fastq -2 input_2.fastq
                 | samtools sort -@ "\${GALAXY_SLOTS:-4}" -T tmp -O bam -o aligned.bam 2>&1 &&
             samtools view -@ "\${GALAXY_SLOTS:-4}" -f 3 -b -q 38 aligned.bam -o unique.bam &&
-            samtools view -@ "\${GALAXY_SLOTS:-4}" -f 3 -b aligned.bam \
-                | samtools view -@ "\${GALAXY_SLOTS:-4}" -U -b -q 38 - \
-                | samtools sort -@ "\${GALAXY_SLOTS:-4}" -n - -
+            samtools view -@ "\${GALAXY_SLOTS:-4}" -f 3 -b aligned.bam
+                | samtools view -@ "\${GALAXY_SLOTS:-4}" -U filt_aligned.bam -bq 38 - > /dev/null &&
+            samtools sort -@ "\${GALAXY_SLOTS:-4}" -T tmp -n filt_aligned.bam
                 | bedtools bamtofastq -i /dev/stdin -fq multimap_1.fastq -fq2 multimap_2.fastq &&
         #end if
         samtools index unique.bam &&

diff --git a/tools/repenrich2/test-data/chrY_paired_class_fraction_counts.tab b/tools/repenrich2/test-data/chrY_paired_class_fraction_counts.tab
@@ -1,6 +1,6 @@
-DNA	148.0
-LINE	195.0
-LTR	26742.0
-Low_complexity	0.0
-RC	0.0
-Simple_repeat	0.0
+DNA	194.66
+LINE	279.83
+LTR	27051.62
+Low_complexity	8.48
+RC	4.24
+Simple_repeat	161.17
diff --git a/tools/repenrich2/test-data/chrY_paired_family_fraction_counts.tab b/tools/repenrich2/test-data/chrY_paired_family_fraction_counts.tab
@@ -1,13 +1,13 @@
-CMC-Transib	30.0
-CR1	4.0
-Copia	25834.0
-Gypsy	857.0
-Helitron	0.0
-Jockey	105.0
-LOA	0.0
-Low_complexity	0.0
-P	32.0
-Pao	51.0
-R1	86.0
-Simple_repeat	0.0
-TcMar-Tc1	86.0
+CMC-Transib	38.48
+CR1	8.24
+Copia	25859.45
+Gypsy	1107.24
+Helitron	4.24
+Jockey	164.38
+LOA	4.24
+Low_complexity	8.48
+P	40.48
+Pao	84.93
+R1	102.97
+Simple_repeat	161.17
+TcMar-Tc1	115.69
diff --git a/tools/repenrich2/test-data/chrY_paired_fraction_counts.tab b/tools/repenrich2/test-data/chrY_paired_fraction_counts.tab
@@ -1,145 +1,145 @@
-A-rich	Low_complexity	Low_complexity	0.0
-ACCORD2_I-int	LTR	Gypsy	0.0
-ACCORD2_LTR	LTR	Gypsy	0.0
-ACCORD_I-int	LTR	Gypsy	0.0
-BARI1	DNA	TcMar-Tc1	0.0
-BATUMI_LTR	LTR	Pao	0.0
-BS	LINE	Jockey	0.0
-BS2	LINE	Jockey	58.0
-BURDOCK_I-int	LTR	Gypsy	0.0
-Baggins1	LINE	LOA	0.0
-Bica_I-int	LTR	Gypsy	38.0
-Bica_LTR	LTR	Gypsy	0.0
-CIRCE	LTR	Gypsy	0.0
-Chouto_I-int	LTR	Gypsy	0.0
-Copia1-I_DM	LTR	Copia	0.0
-Copia_I-int	LTR	Copia	25181.0
-Copia_LTR	LTR	Copia	647.0
-DIVER2_I-int	LTR	Pao	0.0
-DIVER2_LTR	LTR	Pao	0.0
-DM1731_I-int	LTR	Copia	6.0
-DM1731_LTR	LTR	Copia	0.0
-DM176_I-int	LTR	Gypsy	0.0
-DM412	LTR	Gypsy	0.0
-DM412B_LTR	LTR	Gypsy	0.0
-DMCR1A	LINE	CR1	4.0
-DMLTR5	LTR	Gypsy	0.0
-DMRT1A	LINE	R1	0.0
-DMRT1B	LINE	R1	86.0
-DMRT1C	LINE	R1	0.0
-DNAREP1_DM	RC	Helitron	0.0
-DOC2_DM	LINE	Jockey	0.0
-DOC3_DM	LINE	Jockey	0.0
-FB4_DM	DNA	TcMar-Tc1	38.0
-FROGGER_I-int	LTR	Copia	0.0
-FW2_DM	LINE	Jockey	0.0
-G3_DM	LINE	Jockey	0.0
-G5A_DM	LINE	Jockey	0.0
-G5_DM	LINE	Jockey	0.0
-G6_DM	LINE	Jockey	0.0
-GA-rich	Low_complexity	Low_complexity	0.0
-GTWIN_I-int	LTR	Gypsy	0.0
-G_DM	LINE	Jockey	0.0
-Gypsy11_I-int	LTR	Gypsy	0.0
-Gypsy11_LTR	LTR	Gypsy	0.0
-Gypsy12_LTR	LTR	Gypsy	0.0
-Gypsy2-I_DM	LTR	Gypsy	2.0
-Gypsy2-LTR_DM	LTR	Gypsy	0.0
-Gypsy3_LTR	LTR	Gypsy	0.0
-Gypsy4_I-int	LTR	Gypsy	0.0
-Gypsy5_I-int	LTR	Gypsy	0.0
-Gypsy6A_LTR	LTR	Gypsy	0.0
-Gypsy6_I-int	LTR	Gypsy	26.0
-Gypsy8_I-int	LTR	Gypsy	0.0
-Gypsy8_LTR	LTR	Gypsy	0.0
-Gypsy9_I-int	LTR	Gypsy	0.0
-Gypsy_I-int	LTR	Gypsy	32.0
-Gypsy_LTR	LTR	Gypsy	1.0
-HELENA_RT	LINE	Jockey	0.0
-HETA	LINE	Jockey	24.0
-HMSBEAGLE_I-int	LTR	Gypsy	2.0
-IDEFIX_I-int	LTR	Gypsy	4.0
-IDEFIX_LTR	LTR	Gypsy	0.0
-Invader1_I-int	LTR	Gypsy	0.0
-Invader1_LTR	LTR	Gypsy	0.0
-Invader2_I-int	LTR	Gypsy	0.0
-Invader4_I-int	LTR	Gypsy	0.0
-Invader4_LTR	LTR	Gypsy	0.0
-Invader5_I-int	LTR	Gypsy	0.0
-Invader5_LTR	LTR	Gypsy	0.0
-Invader6_I-int	LTR	Gypsy	0.0
-Invader6_LTR	LTR	Gypsy	0.0
-MAX_I-int	LTR	Pao	49.0
-MAX_LTR	LTR	Pao	2.0
-MDG1_I-int	LTR	Gypsy	0.0
-MDG1_LTR	LTR	Gypsy	0.0
-MDG3_I-int	LTR	Gypsy	152.0
-MDG3_LTR	LTR	Gypsy	0.0
-MICROPIA_I-int	LTR	Gypsy	0.0
-MICROPIA_LTR	LTR	Gypsy	0.0
-Mariner2_DM	DNA	TcMar-Tc1	0.0
-NINJA_I-int	LTR	Pao	0.0
-NOMAD_I-int	LTR	Gypsy	0.0
-PROTOP_A	DNA	P	32.0
-PROTOP_B	DNA	P	0.0
-QUASIMODO2-I_DM	LTR	Gypsy	42.0
-QUASIMODO2-LTR_DM	LTR	Gypsy	0.0
-QUASIMODO_I-int	LTR	Gypsy	10.0
-QUASIMODO_LTR	LTR	Gypsy	2.0
-R1_DM	LINE	R1	0.0
-ROOA_I-int	LTR	Pao	0.0
-ROOA_LTR	LTR	Pao	0.0
-ROVER-I_DM	LTR	Gypsy	381.0
-ROVER-LTR_DM	LTR	Gypsy	2.0
-S2_DM	DNA	TcMar-Tc1	0.0
-STALKER4_I-int	LTR	Gypsy	77.0
-STALKER4_LTR	LTR	Gypsy	4.0
-S_DM	DNA	TcMar-Tc1	48.0
-Stalker2_I-int	LTR	Gypsy	80.0
-Stalker2_LTR	LTR	Gypsy	2.0
-TART-A	LINE	Jockey	4.0
-TART_B1	LINE	Jockey	19.0
-TC1-2_DM	DNA	TcMar-Tc1	0.0
-TC1_DM	DNA	TcMar-Tc1	0.0
-TLD2	LTR	Gypsy	0.0
-TRANSIB1	DNA	CMC-Transib	0.0
-TRANSIB2	DNA	CMC-Transib	30.0
-ZAM_I-int	LTR	Gypsy	0.0
-_AACACA_n	Simple_repeat	Simple_repeat	0.0
-_AAT_n	Simple_repeat	Simple_repeat	0.0
-_ACAATAG_n	Simple_repeat	Simple_repeat	0.0
-_ACC_n	Simple_repeat	Simple_repeat	0.0
-_AGAGAAG_n	Simple_repeat	Simple_repeat	0.0
-_AGAGA_n	Simple_repeat	Simple_repeat	0.0
-_ATAAT_n	Simple_repeat	Simple_repeat	0.0
-_ATATATT_n	Simple_repeat	Simple_repeat	0.0
-_ATATTAT_n	Simple_repeat	Simple_repeat	0.0
-_ATTTTT_n	Simple_repeat	Simple_repeat	0.0
-_ATT_n	Simple_repeat	Simple_repeat	0.0
-_AT_n	Simple_repeat	Simple_repeat	0.0
-_A_n	Simple_repeat	Simple_repeat	0.0
-_CATA_n	Simple_repeat	Simple_repeat	0.0
-_CTTTT_n	Simple_repeat	Simple_repeat	0.0
-_GAGAA_n	Simple_repeat	Simple_repeat	0.0
-_GCCTTT_n	Simple_repeat	Simple_repeat	0.0
-_TAATAT_n	Simple_repeat	Simple_repeat	0.0
-_TAATA_n	Simple_repeat	Simple_repeat	0.0
-_TATAAAA_n	Simple_repeat	Simple_repeat	0.0
-_TATAA_n	Simple_repeat	Simple_repeat	0.0
-_TATCATG_n	Simple_repeat	Simple_repeat	0.0
-_TA_n	Simple_repeat	Simple_repeat	0.0
-_TGTTG_n	Simple_repeat	Simple_repeat	0.0
-_TTATATA_n	Simple_repeat	Simple_repeat	0.0
-_TTATAT_n	Simple_repeat	Simple_repeat	0.0
-_TTATA_n	Simple_repeat	Simple_repeat	0.0
-_TTA_n	Simple_repeat	Simple_repeat	0.0
-_TTCTT_n	Simple_repeat	Simple_repeat	0.0
-_TTC_n	Simple_repeat	Simple_repeat	0.0
-_TTTAT_n	Simple_repeat	Simple_repeat	0.0
-_TTTA_n	Simple_repeat	Simple_repeat	0.0
-_TTTC_n	Simple_repeat	Simple_repeat	0.0
-_TTTGA_n	Simple_repeat	Simple_repeat	0.0
-_TTTTAG_n	Simple_repeat	Simple_repeat	0.0
-_TTTTCTT_n	Simple_repeat	Simple_repeat	0.0
-_TTTTC_n	Simple_repeat	Simple_repeat	0.0
-_T_n	Simple_repeat	Simple_repeat	0.0
+A-rich	Low_complexity	Low_complexity	4.24
+ACCORD2_I-int	LTR	Gypsy	4.24
+ACCORD2_LTR	LTR	Gypsy	4.24
+ACCORD_I-int	LTR	Gypsy	4.24
+BARI1	DNA	TcMar-Tc1	4.24
+BATUMI_LTR	LTR	Pao	4.24
+BS	LINE	Jockey	4.24
+BS2	LINE	Jockey	62.24
+BURDOCK_I-int	LTR	Gypsy	4.24
+Baggins1	LINE	LOA	4.24
+Bica_I-int	LTR	Gypsy	42.24
+Bica_LTR	LTR	Gypsy	4.24
+CIRCE	LTR	Gypsy	4.24
+Chouto_I-int	LTR	Gypsy	4.24
+Copia1-I_DM	LTR	Copia	4.24
+Copia_I-int	LTR	Copia	25185.24
+Copia_LTR	LTR	Copia	651.24
+DIVER2_I-int	LTR	Pao	4.24
+DIVER2_LTR	LTR	Pao	4.24
+DM1731_I-int	LTR	Copia	10.24
+DM1731_LTR	LTR	Copia	4.24
+DM176_I-int	LTR	Gypsy	4.24
+DM412	LTR	Gypsy	4.24
+DM412B_LTR	LTR	Gypsy	4.24
+DMCR1A	LINE	CR1	8.24
+DMLTR5	LTR	Gypsy	4.24
+DMRT1A	LINE	R1	4.24
+DMRT1B	LINE	R1	90.24
+DMRT1C	LINE	R1	4.24
+DNAREP1_DM	RC	Helitron	4.24
+DOC2_DM	LINE	Jockey	4.24
+DOC3_DM	LINE	Jockey	4.24
+FB4_DM	DNA	TcMar-Tc1	42.24
+FROGGER_I-int	LTR	Copia	4.24
+FW2_DM	LINE	Jockey	4.24
+G3_DM	LINE	Jockey	4.24
+G5A_DM	LINE	Jockey	4.24
+G5_DM	LINE	Jockey	4.24
+G6_DM	LINE	Jockey	4.24
+GA-rich	Low_complexity	Low_complexity	4.24
+GTWIN_I-int	LTR	Gypsy	4.24
+G_DM	LINE	Jockey	4.24
+Gypsy11_I-int	LTR	Gypsy	4.24
+Gypsy11_LTR	LTR	Gypsy	4.24
+Gypsy12_LTR	LTR	Gypsy	4.24
+Gypsy2-I_DM	LTR	Gypsy	6.24
+Gypsy2-LTR_DM	LTR	Gypsy	4.24
+Gypsy3_LTR	LTR	Gypsy	4.24
+Gypsy4_I-int	LTR	Gypsy	4.24
+Gypsy5_I-int	LTR	Gypsy	4.24
+Gypsy6A_LTR	LTR	Gypsy	4.24
+Gypsy6_I-int	LTR	Gypsy	30.24
+Gypsy8_I-int	LTR	Gypsy	4.24
+Gypsy8_LTR	LTR	Gypsy	4.24
+Gypsy9_I-int	LTR	Gypsy	4.24
+Gypsy_I-int	LTR	Gypsy	36.24
+Gypsy_LTR	LTR	Gypsy	5.24
+HELENA_RT	LINE	Jockey	4.24
+HETA	LINE	Jockey	28.24
+HMSBEAGLE_I-int	LTR	Gypsy	6.24
+IDEFIX_I-int	LTR	Gypsy	8.24
+IDEFIX_LTR	LTR	Gypsy	4.24
+Invader1_I-int	LTR	Gypsy	4.24
+Invader1_LTR	LTR	Gypsy	4.24
+Invader2_I-int	LTR	Gypsy	4.24
+Invader4_I-int	LTR	Gypsy	4.24
+Invader4_LTR	LTR	Gypsy	4.24
+Invader5_I-int	LTR	Gypsy	4.24
+Invader5_LTR	LTR	Gypsy	4.24
+Invader6_I-int	LTR	Gypsy	4.24
+Invader6_LTR	LTR	Gypsy	4.24
+MAX_I-int	LTR	Pao	53.24
+MAX_LTR	LTR	Pao	6.24
+MDG1_I-int	LTR	Gypsy	4.24
+MDG1_LTR	LTR	Gypsy	4.24
+MDG3_I-int	LTR	Gypsy	156.24
+MDG3_LTR	LTR	Gypsy	4.24
+MICROPIA_I-int	LTR	Gypsy	4.24
+MICROPIA_LTR	LTR	Gypsy	4.24
+Mariner2_DM	DNA	TcMar-Tc1	4.24
+NINJA_I-int	LTR	Pao	4.24
+NOMAD_I-int	LTR	Gypsy	4.24
+PROTOP_A	DNA	P	36.24
+PROTOP_B	DNA	P	4.24
+QUASIMODO2-I_DM	LTR	Gypsy	46.24
+QUASIMODO2-LTR_DM	LTR	Gypsy	4.24
+QUASIMODO_I-int	LTR	Gypsy	14.24
+QUASIMODO_LTR	LTR	Gypsy	6.24
+R1_DM	LINE	R1	4.24
+ROOA_I-int	LTR	Pao	4.24
+ROOA_LTR	LTR	Pao	4.24
+ROVER-I_DM	LTR	Gypsy	385.24
+ROVER-LTR_DM	LTR	Gypsy	6.24
+S2_DM	DNA	TcMar-Tc1	4.24
+STALKER4_I-int	LTR	Gypsy	81.24
+STALKER4_LTR	LTR	Gypsy	8.24
+S_DM	DNA	TcMar-Tc1	52.24
+Stalker2_I-int	LTR	Gypsy	84.24
+Stalker2_LTR	LTR	Gypsy	6.24
+TART-A	LINE	Jockey	8.24
+TART_B1	LINE	Jockey	23.24
+TC1-2_DM	DNA	TcMar-Tc1	4.24
+TC1_DM	DNA	TcMar-Tc1	4.24
+TLD2	LTR	Gypsy	4.24
+TRANSIB1	DNA	CMC-Transib	4.24
+TRANSIB2	DNA	CMC-Transib	34.24
+ZAM_I-int	LTR	Gypsy	4.24
+_AACACA_n	Simple_repeat	Simple_repeat	4.24
+_AAT_n	Simple_repeat	Simple_repeat	4.24
+_ACAATAG_n	Simple_repeat	Simple_repeat	4.24
+_ACC_n	Simple_repeat	Simple_repeat	4.24
+_AGAGAAG_n	Simple_repeat	Simple_repeat	4.24
+_AGAGA_n	Simple_repeat	Simple_repeat	4.24
+_ATAAT_n	Simple_repeat	Simple_repeat	4.24
+_ATATATT_n	Simple_repeat	Simple_repeat	4.24
+_ATATTAT_n	Simple_repeat	Simple_repeat	4.24
+_ATTTTT_n	Simple_repeat	Simple_repeat	4.24
+_ATT_n	Simple_repeat	Simple_repeat	4.24
+_AT_n	Simple_repeat	Simple_repeat	4.24
+_A_n	Simple_repeat	Simple_repeat	4.24
+_CATA_n	Simple_repeat	Simple_repeat	4.24
+_CTTTT_n	Simple_repeat	Simple_repeat	4.24
+_GAGAA_n	Simple_repeat	Simple_repeat	4.24
+_GCCTTT_n	Simple_repeat	Simple_repeat	4.24
+_TAATAT_n	Simple_repeat	Simple_repeat	4.24
+_TAATA_n	Simple_repeat	Simple_repeat	4.24
+_TATAAAA_n	Simple_repeat	Simple_repeat	4.24
+_TATAA_n	Simple_repeat	Simple_repeat	4.24
+_TATCATG_n	Simple_repeat	Simple_repeat	4.24
+_TA_n	Simple_repeat	Simple_repeat	4.24
+_TGTTG_n	Simple_repeat	Simple_repeat	4.24
+_TTATATA_n	Simple_repeat	Simple_repeat	4.24
+_TTATAT_n	Simple_repeat	Simple_repeat	4.24
+_TTATA_n	Simple_repeat	Simple_repeat	4.24
+_TTA_n	Simple_repeat	Simple_repeat	4.24
+_TTCTT_n	Simple_repeat	Simple_repeat	4.24
+_TTC_n	Simple_repeat	Simple_repeat	4.24
+_TTTAT_n	Simple_repeat	Simple_repeat	4.24
+_TTTA_n	Simple_repeat	Simple_repeat	4.24
+_TTTC_n	Simple_repeat	Simple_repeat	4.24
+_TTTGA_n	Simple_repeat	Simple_repeat	4.24
+_TTTTAG_n	Simple_repeat	Simple_repeat	4.24
+_TTTTCTT_n	Simple_repeat	Simple_repeat	4.24
+_TTTTC_n	Simple_repeat	Simple_repeat	4.24
+_T_n	Simple_repeat	Simple_repeat	4.24
diff --git a/tools/repenrich2/test-data/chrY_single_class_fraction_counts.tab b/tools/repenrich2/test-data/chrY_single_class_fraction_counts.tab
@@ -1,6 +1,6 @@
-DNA	70.0
-LINE	90.0
-LTR	13210.0
-Low_complexity	0.0
-RC	0.0
-Simple_repeat	0.0
+DNA	121.89
+LINE	184.34
+LTR	13554.36
+Low_complexity	9.43
+RC	4.72
+Simple_repeat	179.26