# HG changeset patch
# User jackcurragh
# Date 1648039895 0
# Node ID 5bdc8418c3785825307c28dcc82ddcf250316c81

Uploaded

diff -r 000000000000 -r 5bdc8418c378 samtools_index/macros.xml
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/samtools_index/macros.xml	Wed Mar 23 12:51:35 2022 +0000
@@ -0,0 +1,223 @@
+<macros>
+    <xml name="requirements">
+        <requirements>
+            <requirement type="package" version="@TOOL_VERSION@">samtools</requirement>
+            <yield/>
+        </requirements>
+    </xml>
+    <token name="@TOOL_VERSION@">1.13</token>
+    <token name="@PROFILE@">20.05</token>
+    <token name="@FLAGS@"><![CDATA[
+        #set $flags = 0
+        #if $filter
+            #set $flags = sum(map(int, str($filter).split(',')))
+        #end if
+    ]]></token>
+    <token name="@PREPARE_IDX@"><![CDATA[
+        ##prepare input and indices
+        ln -s '$input' infile &&
+        #if $input.is_of_type('bam'):
+            #if str( $input.metadata.bam_index ) != "None":
+                ln -s '${input.metadata.bam_index}' infile.bai &&
+            #else:
+                samtools index infile infile.bai &&
+            #end if
+        #elif $input.is_of_type('cram'):
+            #if str( $input.metadata.cram_index ) != "None":
+                ln -s '${input.metadata.cram_index}' infile.crai &&
+            #else:
+                samtools index infile infile.crai &&
+            #end if
+        #end if
+    ]]></token>
+    <token name="@PREPARE_IDX_MULTIPLE@"><![CDATA[
+        ##prepare input and indices
+        #for $i, $bam in enumerate( $input_bams ):
+            ln -s '$bam' '${i}' &&
+            #if $bam.is_of_type('bam'):
+                #if str( $bam.metadata.bam_index ) != "None":
+                    ln -s '${bam.metadata.bam_index}' '${i}.bai' &&
+                #else:
+                    samtools index '${i}' '${i}.bai' &&
+                #end if
+            #elif $bam.is_of_type('cram'):
+                #if str( $bam.metadata.cram_index ) != "None":
+                    ln -s '${bam.metadata.cram_index}' '${i}.crai' &&
+                #else:
+                    samtools index '${i}' '${i}.crai' &&
+                #end if
+            #end if
+        #end for
+    ]]></token>
+    <token name="@PREPARE_FASTA_IDX@"><![CDATA[
+        ##checks for reference data ($addref_cond.addref_select=="history" or =="cached")
+        ##and sets the -t/-T parameters accordingly:
+        ##- in case of history a symbolic link is used because samtools (view) will generate
+        ##  the index which might not be possible in the directory containing the fasta file
+        ##- in case of cached the absolute path is used which allows to read the cram file
+        ##  without specifying the reference
+        #if $addref_cond.addref_select == "history":
+            ln -s '${addref_cond.ref}' reference.fa &&
+            samtools faidx reference.fa &&
+            #set reffa="reference.fa"
+            #set reffai="reference.fa.fai"
+        #elif $addref_cond.addref_select == "cached":
+            #set reffa=str($addref_cond.ref.fields.path)
+            #set reffai=str($addref_cond.ref.fields.path)+".fai"
+        #else
+            #set reffa=None
+            #set reffai=None
+        #end if
+    ]]></token>
+
+    <xml name="optional_reference">
+        <conditional name="addref_cond">
+            <param name="addref_select" type="select" label="Use a reference sequence">
+                <help>@HELP@</help>
+                <option value="no">No</option>
+                <option value="history">Use a genome/index from the history</option>
+                <option value="cached">Use a built-in genome</option>
+            </param>
+            <when value="no"/>
+            <when value="history">
+                <param name="ref" argument="@ARGUMENT@" type="data" format="fasta,fasta.gz" label="Reference"/>
+            </when>
+            <when value="cached">
+                <param name="ref" argument="@ARGUMENT@" type="select" label="Reference">
+                    <options from_data_table="fasta_indexes">
+                        <filter type="data_meta" ref="input" key="dbkey" column="dbkey"/>
+                    </options>
+                    <validator  type="no_options" message="No reference genome is available for the build associated with the selected input dataset"/>
+                </param>
+            </when>
+        </conditional>
+    </xml>
+    <xml name="mandatory_reference" token_help="" token_argument="">
+        <conditional name="addref_cond">
+            <param name="addref_select" type="select" label="Use a reference sequence">
+                <help>@HELP@</help>
+                <option value="history">Use a genome/index from the history</option>
+                <option value="cached">Use a built-in genome</option>
+            </param>
+            <when value="history">
+                <param name="ref" argument="@ARGUMENT@" type="data" format="fasta,fasta.gz" label="Reference"/>
+            </when>
+            <when value="cached">
+                <param name="ref" argument="@ARGUMENT@" type="select" label="Reference">
+                    <options from_data_table="fasta_indexes">
+                        <filter type="data_meta" ref="input" key="dbkey" column="dbkey"/>
+                        <validator message="No reference genome is available for the build associated with the selected input dataset" type="no_options" />
+                    </options>
+                </param>
+            </when>
+        </conditional>
+    </xml>
+
+
+    <token name="@ADDTHREADS@"><![CDATA[
+        ##compute the number of ADDITIONAL threads to be used by samtools (-@)
+        addthreads=\${GALAXY_SLOTS:-1} && (( addthreads-- )) &&
+    ]]></token>
+    <token name="@ADDMEMORY@"><![CDATA[
+        ##compute the number of memory available to samtools sort (-m)
+        ##use only 75% of available: https://github.com/samtools/samtools/issues/831
+        addmemory=\${GALAXY_MEMORY_MB_PER_SLOT:-768} &&
+        ((addmemory=addmemory*75/100)) &&
+    ]]></token>
+    <xml name="seed_input">
+       <param name="seed" type="integer" optional="True" label="Seed for random number generator" help="If empty a random seed is used." />
+    </xml>
+    <xml name="flag_options" token_s1="false" token_s2="false" token_s4="false" token_s8="false" token_s16="false" token_s32="false" token_s64="false" token_s128="false" token_s256="false" token_s512="false" token_s1024="false" token_s2048="false">
+        <option value="1" selected="@S1@">Read is paired</option>
+        <option value="2" selected="@S2@">Read is mapped in a proper pair</option>
+        <option value="4" selected="@S4@">Read is unmapped</option>
+        <option value="8" selected="@S8@">Mate is unmapped</option>
+        <option value="16" selected="@S16@">Read is mapped to the reverse strand of the reference</option>
+        <option value="32" selected="@S32@">Mate is mapped to the reverse strand of the reference</option>
+        <option value="64" selected="@S64@">Read is the first in a pair</option>
+        <option value="128" selected="@S128@">Read is the second in a pair</option>
+        <option value="256" selected="@S256@">Alignment of the read is not primary</option>
+        <option value="512" selected="@S512@">Read fails platform/vendor quality checks</option>
+        <option value="1024" selected="@S1024@">Read is a PCR or optical duplicate</option>
+        <option value="2048" selected="@S2048@">Alignment is supplementary</option>
+    </xml>
+
+    <!-- region specification macros and tokens for tools that allow the specification
+         of region by bed file / space separated list of regions -->
+    <token name="@REGIONS_FILE@"><![CDATA[
+        #if $cond_region.select_region == 'tab':
+            -t '$cond_region.targetregions'
+        #end if
+    ]]></token>
+    <token name="@REGIONS_MANUAL@"><![CDATA[
+        #if $cond_region.select_region == 'text':
+            #for $i, $x in enumerate($cond_region.regions_repeat):
+               '${x.region}'
+            #end for
+        #end if
+    ]]></token>
+    <xml name="regions_macro">
+        <conditional name="cond_region">
+            <param name="select_region" type="select" label="Filter by regions" help="restricts output to only those alignments which overlap the specified region(s)">
+                <option value="no" selected="True">No</option>
+                <option value="text">Manualy specify regions</option>
+                <option value="tab">Regions from tabular file</option>
+            </param>
+            <when value="no"/>
+            <when value="text">
+                <repeat name="regions_repeat" min="1" default="1" title="Regions">
+                    <param name="region" type="text" label="region" help="format chr:from-to">
+                        <validator type="regex" message="Required format: CHR[:FROM[-TO]]; where CHR: string containing any character except quotes, whitespace and colon; FROM and TO: any integer">^[^\s'\":]+(:\d+(-\d+){0,1}){0,1}$</validator>
+                    </param>
+                </repeat>
+            </when>
+            <when value="tab">
+                <param name="targetregions" argument="-t/--target-regions" type="data" format="tabular" label="Target regions file" help="Do stats in these regions only. Tab-delimited file chr,from,to (1-based, inclusive)" />
+            </when>
+        </conditional>
+    </xml>
+
+    <xml name="citations">
+        <citations>
+            <citation type="bibtex">
+                @misc{SAM_def,
+                title={Definition of SAM/BAM format},
+                url = {https://samtools.github.io/hts-specs/},}
+            </citation>
+            <citation type="doi">10.1093/bioinformatics/btp352</citation>
+            <citation type="doi">10.1093/bioinformatics/btr076</citation>
+            <citation type="doi">10.1093/bioinformatics/btr509</citation>
+            <citation type="bibtex">
+                @misc{Danecek_et_al,
+                Author={Danecek, P., Schiffels, S., Durbin, R.},
+                title={Multiallelic calling model in bcftools (-m)},
+                url = {http://samtools.github.io/bcftools/call-m.pdf},}
+            </citation>
+            <citation type="bibtex">
+                @misc{Durbin_VCQC,
+                Author={Durbin, R.},
+                title={Segregation based metric for variant call QC},
+                url = {http://samtools.github.io/bcftools/rd-SegBias.pdf},}
+            </citation>
+            <citation type="bibtex">
+                @misc{Li_SamMath,
+                Author={Li, H.},
+                title={Mathematical Notes on SAMtools Algorithms},
+                url = {http://www.broadinstitute.org/gatk/media/docs/Samtools.pdf},}
+            </citation>
+            <citation type="bibtex">
+                @misc{SamTools_github,
+                title={SAMTools GitHub page},
+                url = {https://github.com/samtools/samtools},}
+            </citation>
+        </citations>
+    </xml>
+    <xml name="version_command">
+        <version_command><![CDATA[samtools 2>&1 | grep Version]]></version_command>
+    </xml>
+    <xml name="stdio">
+        <stdio>
+            <exit_code range="1:" level="fatal" description="Error" />
+        </stdio>
+    </xml>
+</macros>
diff -r 000000000000 -r 5bdc8418c378 samtools_index/samtools_index.xml
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/samtools_index/samtools_index.xml	Wed Mar 23 12:51:35 2022 +0000
@@ -0,0 +1,51 @@
+<tool id="samtools_index" name="Samtools index" version="2.0.4" profile="@PROFILE@">
+    <description>order of storing aligned sequences</description>
+    <macros>
+        <import>macros.xml</import>
+    </macros>
+    <expand macro="requirements"/>
+    <expand macro="stdio"/>
+    <expand macro="version_command"/>
+    <command><![CDATA[
+        @ADDTHREADS@
+	@ADDMEMORY@
+        samtools index
+            -@ \$addthreads
+            -m \$addmemory"M"
+            $prim_key_cond.prim_key_select
+            '${input1}'
+            '${output1}'
+    ]]></command>
+    <inputs>
+        <param name="input1" type="data" format="sam,unsorted.bam,cram" label="BAM File" />
+        <conditional name="prim_key_cond">
+            <param name="prim_key_select" type="select" label="Index type">
+                <option value="-b" selected="True">BAI (-b)</option>
+                <option value="-c" selected="True">CSI (-c)</option>
+            </param>
+            <when value="-b"/>
+            <when value="-c"/>
+        </conditional>
+    </inputs>
+    <outputs>
+       <data name="output1" format="bam">
+            <change_format>
+                <when input="prim_key_cond.prim_key_select" value="-b" format="bai" />
+                <when input="prim_key_cond.prim_key_select" value="-c" format="csi" />
+            </change_format>
+        </data>
+    </outputs>
+    <tests>
+        <test>
+            <param name="input1" value="1.bam" ftype="bam" />
+            <param name="prim_key_cond.prim_key_select" value="-c" />
+            <output name="output1" file="1.bam.bai" ftype="bai" lines_diff="4" />
+        </test>
+    </tests>
+    <help>
+**What it does**
+
+Indexes a BAM file 
+    </help>
+    <expand macro="citations"/>
+</tool>
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/1.bam
Binary file samtools_index/test-data/1.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/1.bam.bai
Binary file samtools_index/test-data/1.bam.bai has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/1_sort.bam
Binary file samtools_index/test-data/1_sort.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/1_sort_read_names.bam
Binary file samtools_index/test-data/1_sort_read_names.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/minhash.expected.bam
Binary file samtools_index/test-data/minhash.expected.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/name.sort.expected.bam
Binary file samtools_index/test-data/name.sort.expected.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/name.sort.expected.sam
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/samtools_index/test-data/name.sort.expected.sam	Wed Mar 23 12:51:35 2022 +0000
@@ -0,0 +1,28 @@
+@HD	VN:1.4	SO:queryname
+@SQ	SN:insert	LN:599
+@SQ	SN:ref1	LN:45
+@SQ	SN:ref2	LN:40
+@SQ	SN:ref3	LN:4
+@RG	ID:fish	PG:donkey
+@RG	ID:cow	PU:13_&^&&*(:332
+@RG	PU:*9u8jkjjkjd:	ID:colt
+@PG	ID:bull	PP:donkey
+@PG	ID:donkey
+@PG	ID:moose
+@PG	PP:moose	ID:cow
+@CO	
+r000	99	insert	50	30	10M	=	80	30	ATTTAGCTAC	AAAAAAAAAA	RG:Z:cow	PG:Z:bull
+r000	211	insert	80	30	10M	=	50	-30	CCCAATCATT	AAAAAAAAAA	RG:Z:cow	PG:Z:bull
+r001	83	ref1	37	30	9M	=	7	-39	CAGCGCCAT	*	RG:Z:fish	PG:Z:colt
+r001	163	ref1	7	30	8M4I4M1D3M	=	37	39	TTAGATAAAGAGGATACTG	*	XX:B:S,12561,2,20,112	YY:i:100	RG:Z:fish	PG:Z:colt
+r002	0	ref1	9	30	1S2I6M1P1I1P1I4M2I	*	0	0	AAAAGATAAGGGATAAA	*	XA:Z:abc	XB:i:-10	PG:Z:colt
+r003	0	ref1	9	30	5H6M	*	0	0	AGCTAA	*	RG:Z:cow
+r003	16	ref1	29	30	6H5M	*	0	0	TAGGC	*	RG:Z:cow	PG:Z:colt
+r004	0	ref1	16	30	6M14N1I5M	*	0	0	ATAGCTCTCAGC	*	RG:Z:colt	PG:Z:colt
+u1	4	*	0	30	23M	*	0	0	TAATTAAGTCTACAGAAAAAAAA	???????????????????????
+x1	0	ref2	1	30	20M	*	0	0	AGGTTTTATAAAACAAATAA	*	RG:Z:colt	PG:Z:bull
+x2	0	ref2	2	30	21M	*	0	0	GGTTTTATAAAACAAATAATT	?????????????????????	RG:Z:colt	PG:Z:bull
+x3	0	ref2	6	30	9M4I13M	*	0	0	TTATAAAACAAATAATTAAGTCTACA	??????????????????????????	RG:Z:fish	PG:Z:bull
+x4	0	ref2	10	30	25M	*	0	0	CAAATAATTAAGTCTACAGAGCAAC	?????????????????????????	RG:Z:fish	PG:Z:bull
+x5	0	ref2	12	30	24M	*	0	0	AATAATTAAGTCTACAGAGCAACT	????????????????????????	RG:Z:fish	PG:Z:bull
+x6	0	ref2	14	30	23M	*	0	0	TAATTAAGTCTACAGAGCAACTA	???????????????????????	RG:Z:cow
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/pos.sort.expected.bam
Binary file samtools_index/test-data/pos.sort.expected.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/pos.sort.expected.sam
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/samtools_index/test-data/pos.sort.expected.sam	Wed Mar 23 12:51:35 2022 +0000
@@ -0,0 +1,28 @@
+@HD	VN:1.4	SO:coordinate
+@SQ	SN:insert	LN:599
+@SQ	SN:ref1	LN:45
+@SQ	SN:ref2	LN:40
+@SQ	SN:ref3	LN:4
+@RG	ID:fish	PG:donkey
+@RG	ID:cow	PU:13_&^&&*(:332
+@RG	PU:*9u8jkjjkjd:	ID:colt
+@PG	ID:bull	PP:donkey
+@PG	ID:donkey
+@PG	ID:moose
+@PG	PP:moose	ID:cow
+@CO	
+r000	99	insert	50	30	10M	=	80	30	ATTTAGCTAC	AAAAAAAAAA	RG:Z:cow	PG:Z:bull
+r000	211	insert	80	30	10M	=	50	-30	CCCAATCATT	AAAAAAAAAA	RG:Z:cow	PG:Z:bull
+r001	163	ref1	7	30	8M4I4M1D3M	=	37	39	TTAGATAAAGAGGATACTG	*	XX:B:S,12561,2,20,112	YY:i:100	RG:Z:fish	PG:Z:colt
+r002	0	ref1	9	30	1S2I6M1P1I1P1I4M2I	*	0	0	AAAAGATAAGGGATAAA	*	XA:Z:abc	XB:i:-10	PG:Z:colt
+r003	0	ref1	9	30	5H6M	*	0	0	AGCTAA	*	RG:Z:cow
+r004	0	ref1	16	30	6M14N1I5M	*	0	0	ATAGCTCTCAGC	*	RG:Z:colt	PG:Z:colt
+r003	16	ref1	29	30	6H5M	*	0	0	TAGGC	*	RG:Z:cow	PG:Z:colt
+r001	83	ref1	37	30	9M	=	7	-39	CAGCGCCAT	*	RG:Z:fish	PG:Z:colt
+x1	0	ref2	1	30	20M	*	0	0	AGGTTTTATAAAACAAATAA	*	RG:Z:colt	PG:Z:bull
+x2	0	ref2	2	30	21M	*	0	0	GGTTTTATAAAACAAATAATT	?????????????????????	RG:Z:colt	PG:Z:bull
+x3	0	ref2	6	30	9M4I13M	*	0	0	TTATAAAACAAATAATTAAGTCTACA	??????????????????????????	RG:Z:fish	PG:Z:bull
+x4	0	ref2	10	30	25M	*	0	0	CAAATAATTAAGTCTACAGAGCAAC	?????????????????????????	RG:Z:fish	PG:Z:bull
+x5	0	ref2	12	30	24M	*	0	0	AATAATTAAGTCTACAGAGCAACT	????????????????????????	RG:Z:fish	PG:Z:bull
+x6	0	ref2	14	30	23M	*	0	0	TAATTAAGTCTACAGAGCAACTA	???????????????????????	RG:Z:cow
+u1	4	*	0	30	23M	*	0	0	TAATTAAGTCTACAGAAAAAAAA	???????????????????????
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/tag.as.sort.expected.bam
Binary file samtools_index/test-data/tag.as.sort.expected.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/tag.as.sort.expected.sam
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/samtools_index/test-data/tag.as.sort.expected.sam	Wed Mar 23 12:51:35 2022 +0000
@@ -0,0 +1,24 @@
+@HD	VN:1.4	SO:unknown
+@SQ	SN:insert	LN:599
+@SQ	SN:ref1	LN:45
+@SQ	SN:ref2	LN:40
+@SQ	SN:ref3	LN:4
+@PG	ID:llama
+@RG	ID:fish	PG:llama
+@RG	ID:cow	PU:13_&^&&*(:332	PG:donkey
+@RG	PU:*9u8jkjjkjd:	ID:colt
+@PG	ID:bull	PP:donkey
+@PG	ID:donkey
+@CO	Do you know?
+r006	16	ref1	29	30	6H5M	*	0	0	TAGGC	*	RG:Z:colt	PG:Z:donkey	FI:i:3
+x11	0	ref2	12	30	24M	*	0	0	AATAATTAAGTCTACAGAGCAACT	????????????????????????	RG:Z:cow	PG:Z:bull	FI:Z:a
+r007	0	ref1	16	30	6M14N1I5M	*	0	0	ATAGCTCTCAGC	*	RG:Z:colt	PG:Z:donkey	AS:i:-5	FI:f:3.5
+x10	0	ref2	10	30	25M	*	0	0	CAAATAATTAAGTCTACAGAGCAAC	?????????????????????????	RG:Z:cow	PG:Z:bull	AS:i:0	FI:A:b
+r007	0	ref1	9	30	5H6M	*	0	0	AGCTAA	*	RG:Z:colt	PG:Z:donkey	AS:i:1	FI:i:4
+r005	163	ref1	7	30	8M4I4M1D3M	=	37	39	TTAGATAAAGAGGATACTG	*	XX:B:S,12561,2,20,112	YY:i:100	RG:Z:colt	PG:Z:donkey	AS:i:10	FI:i:5
+x8	0	ref2	2	30	21M	*	0	0	GGTTTTATAAAACAAATAATT	?????????????????????	RG:Z:cow	PG:Z:bull	AS:i:10	FI:f:1.5
+r006	0	ref1	9	30	1S2I6M1P1I1P1I4M2I	*	0	0	AAAAGATAAGGGATAAA	*	XA:Z:abc	XB:i:-10	RG:Z:colt	PG:Z:donkey	AS:i:20	FI:f:4.5
+x9	0	ref2	6	30	9M4I13M	*	0	0	TTATAAAACAAATAATTAAGTCTACA	??????????????????????????	RG:Z:cow	PG:Z:bull	AS:i:20	FI:i:1
+x7	0	ref2	1	30	20M	*	0	0	AGGTTTTATAAAACAAATAA	*	RG:Z:cow	PG:Z:bull	AS:i:50	FI:i:2
+r005	83	ref1	37	30	9M	=	7	-39	CAGCGCCAT	*	RG:Z:colt	PG:Z:donkey	AS:i:100	FI:f:2.5
+x12	0	ref2	14	30	23M	*	0	0	TAATTAAGTCTACAGAGCAACTA	???????????????????????	RG:Z:cow	PG:Z:bull	AS:i:65100
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/tag.fi.sort.expected.bam
Binary file samtools_index/test-data/tag.fi.sort.expected.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/tag.fi.sort.expected.sam
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/samtools_index/test-data/tag.fi.sort.expected.sam	Wed Mar 23 12:51:35 2022 +0000
@@ -0,0 +1,24 @@
+@HD	VN:1.4	SO:unknown
+@SQ	SN:insert	LN:599
+@SQ	SN:ref1	LN:45
+@SQ	SN:ref2	LN:40
+@SQ	SN:ref3	LN:4
+@PG	ID:llama
+@RG	ID:fish	PG:llama
+@RG	ID:cow	PU:13_&^&&*(:332	PG:donkey
+@RG	PU:*9u8jkjjkjd:	ID:colt
+@PG	ID:bull	PP:donkey
+@PG	ID:donkey
+@CO	Do you know?
+x12	0	ref2	14	30	23M	*	0	0	TAATTAAGTCTACAGAGCAACTA	???????????????????????	RG:Z:cow	PG:Z:bull	AS:i:65100
+x10	0	ref2	10	30	25M	*	0	0	CAAATAATTAAGTCTACAGAGCAAC	?????????????????????????	RG:Z:cow	PG:Z:bull	AS:i:0	FI:A:b
+x11	0	ref2	12	30	24M	*	0	0	AATAATTAAGTCTACAGAGCAACT	????????????????????????	RG:Z:cow	PG:Z:bull	FI:Z:a
+x9	0	ref2	6	30	9M4I13M	*	0	0	TTATAAAACAAATAATTAAGTCTACA	??????????????????????????	RG:Z:cow	PG:Z:bull	AS:i:20	FI:i:1
+x8	0	ref2	2	30	21M	*	0	0	GGTTTTATAAAACAAATAATT	?????????????????????	RG:Z:cow	PG:Z:bull	AS:i:10	FI:f:1.5
+x7	0	ref2	1	30	20M	*	0	0	AGGTTTTATAAAACAAATAA	*	RG:Z:cow	PG:Z:bull	AS:i:50	FI:i:2
+r005	83	ref1	37	30	9M	=	7	-39	CAGCGCCAT	*	RG:Z:colt	PG:Z:donkey	AS:i:100	FI:f:2.5
+r006	16	ref1	29	30	6H5M	*	0	0	TAGGC	*	RG:Z:colt	PG:Z:donkey	FI:i:3
+r007	0	ref1	16	30	6M14N1I5M	*	0	0	ATAGCTCTCAGC	*	RG:Z:colt	PG:Z:donkey	AS:i:-5	FI:f:3.5
+r007	0	ref1	9	30	5H6M	*	0	0	AGCTAA	*	RG:Z:colt	PG:Z:donkey	AS:i:1	FI:i:4
+r006	0	ref1	9	30	1S2I6M1P1I1P1I4M2I	*	0	0	AAAAGATAAGGGATAAA	*	XA:Z:abc	XB:i:-10	RG:Z:colt	PG:Z:donkey	AS:i:20	FI:f:4.5
+r005	163	ref1	7	30	8M4I4M1D3M	=	37	39	TTAGATAAAGAGGATACTG	*	XX:B:S,12561,2,20,112	YY:i:100	RG:Z:colt	PG:Z:donkey	AS:i:10	FI:i:5
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/tag.rg.n.sort.expected.bam
Binary file samtools_index/test-data/tag.rg.n.sort.expected.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/tag.rg.n.sort.expected.sam
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/samtools_index/test-data/tag.rg.n.sort.expected.sam	Wed Mar 23 12:51:35 2022 +0000
@@ -0,0 +1,28 @@
+@HD	VN:1.4	SO:unknown
+@SQ	SN:insert	LN:599
+@SQ	SN:ref1	LN:45
+@SQ	SN:ref2	LN:40
+@SQ	SN:ref3	LN:4
+@RG	ID:fish	PG:donkey
+@RG	ID:cow	PU:13_&^&&*(:332
+@RG	PU:*9u8jkjjkjd:	ID:colt
+@PG	ID:bull	PP:donkey
+@PG	ID:donkey
+@PG	ID:moose
+@PG	PP:moose	ID:cow
+@CO	
+r002	0	ref1	9	30	1S2I6M1P1I1P1I4M2I	*	0	0	AAAAGATAAGGGATAAA	*	XA:Z:abc	XB:i:-10	PG:Z:colt
+u1	4	*	0	30	23M	*	0	0	TAATTAAGTCTACAGAAAAAAAA	???????????????????????
+r004	0	ref1	16	30	6M14N1I5M	*	0	0	ATAGCTCTCAGC	*	RG:Z:colt	PG:Z:colt
+x1	0	ref2	1	30	20M	*	0	0	AGGTTTTATAAAACAAATAA	*	RG:Z:colt	PG:Z:bull
+x2	0	ref2	2	30	21M	*	0	0	GGTTTTATAAAACAAATAATT	?????????????????????	RG:Z:colt	PG:Z:bull
+r000	99	insert	50	30	10M	=	80	30	ATTTAGCTAC	AAAAAAAAAA	RG:Z:cow	PG:Z:bull
+r000	211	insert	80	30	10M	=	50	-30	CCCAATCATT	AAAAAAAAAA	RG:Z:cow	PG:Z:bull
+r003	0	ref1	9	30	5H6M	*	0	0	AGCTAA	*	RG:Z:cow
+r003	16	ref1	29	30	6H5M	*	0	0	TAGGC	*	RG:Z:cow	PG:Z:colt
+x6	0	ref2	14	30	23M	*	0	0	TAATTAAGTCTACAGAGCAACTA	???????????????????????	RG:Z:cow
+r001	83	ref1	37	30	9M	=	7	-39	CAGCGCCAT	*	RG:Z:fish	PG:Z:colt
+r001	163	ref1	7	30	8M4I4M1D3M	=	37	39	TTAGATAAAGAGGATACTG	*	XX:B:S,12561,2,20,112	YY:i:100	RG:Z:fish	PG:Z:colt
+x3	0	ref2	6	30	9M4I13M	*	0	0	TTATAAAACAAATAATTAAGTCTACA	??????????????????????????	RG:Z:fish	PG:Z:bull
+x4	0	ref2	10	30	25M	*	0	0	CAAATAATTAAGTCTACAGAGCAAC	?????????????????????????	RG:Z:fish	PG:Z:bull
+x5	0	ref2	12	30	24M	*	0	0	AATAATTAAGTCTACAGAGCAACT	????????????????????????	RG:Z:fish	PG:Z:bull
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/tag.rg.sort.expected.bam
Binary file samtools_index/test-data/tag.rg.sort.expected.bam has changed
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/tag.rg.sort.expected.sam
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/samtools_index/test-data/tag.rg.sort.expected.sam	Wed Mar 23 12:51:35 2022 +0000
@@ -0,0 +1,28 @@
+@HD	VN:1.4	SO:unknown
+@SQ	SN:insert	LN:599
+@SQ	SN:ref1	LN:45
+@SQ	SN:ref2	LN:40
+@SQ	SN:ref3	LN:4
+@RG	ID:fish	PG:donkey
+@RG	ID:cow	PU:13_&^&&*(:332
+@RG	PU:*9u8jkjjkjd:	ID:colt
+@PG	ID:bull	PP:donkey
+@PG	ID:donkey
+@PG	ID:moose
+@PG	PP:moose	ID:cow
+@CO	
+r002	0	ref1	9	30	1S2I6M1P1I1P1I4M2I	*	0	0	AAAAGATAAGGGATAAA	*	XA:Z:abc	XB:i:-10	PG:Z:colt
+u1	4	*	0	30	23M	*	0	0	TAATTAAGTCTACAGAAAAAAAA	???????????????????????
+r004	0	ref1	16	30	6M14N1I5M	*	0	0	ATAGCTCTCAGC	*	RG:Z:colt	PG:Z:colt
+x1	0	ref2	1	30	20M	*	0	0	AGGTTTTATAAAACAAATAA	*	RG:Z:colt	PG:Z:bull
+x2	0	ref2	2	30	21M	*	0	0	GGTTTTATAAAACAAATAATT	?????????????????????	RG:Z:colt	PG:Z:bull
+r000	99	insert	50	30	10M	=	80	30	ATTTAGCTAC	AAAAAAAAAA	RG:Z:cow	PG:Z:bull
+r000	211	insert	80	30	10M	=	50	-30	CCCAATCATT	AAAAAAAAAA	RG:Z:cow	PG:Z:bull
+r003	0	ref1	9	30	5H6M	*	0	0	AGCTAA	*	RG:Z:cow
+r003	16	ref1	29	30	6H5M	*	0	0	TAGGC	*	RG:Z:cow	PG:Z:colt
+x6	0	ref2	14	30	23M	*	0	0	TAATTAAGTCTACAGAGCAACTA	???????????????????????	RG:Z:cow
+r001	163	ref1	7	30	8M4I4M1D3M	=	37	39	TTAGATAAAGAGGATACTG	*	XX:B:S,12561,2,20,112	YY:i:100	RG:Z:fish	PG:Z:colt
+r001	83	ref1	37	30	9M	=	7	-39	CAGCGCCAT	*	RG:Z:fish	PG:Z:colt
+x3	0	ref2	6	30	9M4I13M	*	0	0	TTATAAAACAAATAATTAAGTCTACA	??????????????????????????	RG:Z:fish	PG:Z:bull
+x4	0	ref2	10	30	25M	*	0	0	CAAATAATTAAGTCTACAGAGCAAC	?????????????????????????	RG:Z:fish	PG:Z:bull
+x5	0	ref2	12	30	24M	*	0	0	AATAATTAAGTCTACAGAGCAACT	????????????????????????	RG:Z:fish	PG:Z:bull
diff -r 000000000000 -r 5bdc8418c378 samtools_index/test-data/test_input_1_a.bam
Binary file samtools_index/test-data/test_input_1_a.bam has changed