Mercurial > repos > fubar > microsatbedfubar

changeset 4:25c70fff4079 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/fubar2/microsatbed commit d952bc313f408735456747c3d33e09a3170c8f59
author fubar
date Sun, 14 Jul 2024 23:52:21 +0000
parents 36b8c6f7601b
children
files README.md microsatbed.xml test-data/builtinnativetsv_sample tool-data/all_fasta.loc.sample tool_data_table_conf.xml.sample tool_data_table_conf.xml.test
diffstat 6 files changed, 1003 insertions(+), 21 deletions(-) [+]
line wrap: on
line diff
--- a/README.md	Sun Jul 14 03:55:47 2024 +0000
+++ b/README.md	Sun Jul 14 23:52:21 2024 +0000
@@ -1,10 +1,9 @@
 ## microsatellites to bed features
 
-
  **Convert short repetitive sequences to bed features**
 
- Microsatellites refer to short DNA patterns that are repeated.
- A pattern or *motif* can be any combination nucleotides, typically from 1 to 6nt in length.
+ Microsatellites are usually defined as repeated short DNA patterns in an unbroken sequence.
+ A microsatellite pattern or *motif* can be any combination nucleotides, typically from 1 to 6nt in length.
  
  This tool allows microsatellite and related features to be selected from a fasta sequence input file, and output into a single bed track, suitable for viewing in a genome browser such as JBrowse2.
 
@@ -51,4 +50,4 @@
    Pytrf is a lightweight Python C extension for identification of tandem repeats. The pytrf enables to fastly identify both exact or perfect SSRs.
    It also can find generic tandem repeats with any size of motif, such as with maximum motif length of 100 bp. Additionally, it has capability of finding approximate or imperfect tandem repeats*
 
- 
+ 
\ No newline at end of file
--- a/microsatbed.xml	Sun Jul 14 03:55:47 2024 +0000
+++ b/microsatbed.xml	Sun Jul 14 23:52:21 2024 +0000
@@ -1,4 +1,4 @@
-<tool name="strtobed" id="microsatbedfubar" version="1.3.0" profile="22.05">
+<tool name="STR to bed" id="microsatbed" version="1.3.0" profile="22.05">
   <description>Short Tandem Repeats to bed features from fasta</description>
   <requirements>
     <requirement version="3.12.3" type="package">python</requirement>
@@ -11,10 +11,18 @@
    <version_command><![CDATA[python -c "import pytrf; from importlib.metadata import version; print(version('pytrf'))"]]></version_command>
   <command><![CDATA[
   #if $mode_cond.mode == "NATIVE":
-    pytrf findstr -f '$mode_cond.outformat' -o $bed -r $mode_cond.monomin $mode_cond.dimin $mode_cond.trimin $mode_cond.tetramin $mode_cond.pentamin $mode_cond.hexamin '${reference_genome.fasta}' 
+    #if $reference_genome.genome_type_select == "history":
+      pytrf findstr -f '$mode_cond.outformat' -o $bed -r $mode_cond.monomin $mode_cond.dimin $mode_cond.trimin $mode_cond.tetramin $mode_cond.pentamin $mode_cond.hexamin '${reference_genome.fasta}'
+    #else:
+      pytrf findstr -f '$mode_cond.outformat' -o $bed -r $mode_cond.monomin $mode_cond.dimin $mode_cond.trimin $mode_cond.tetramin $mode_cond.pentamin $mode_cond.hexamin '${reference_genome.fasta.fields.path}'
+    #end if
   #else:
     python '${__tool_directory__}/find_str.py'
-    --fasta '${reference_genome.fasta}' 
+    #if $reference_genome.genome_type_select == "history":
+      --fasta '${reference_genome.fasta}'
+    #else:
+      --fasta '${reference_genome.fasta.fields.path}'
+    #end if
     --bed '$bed'
     #if $mode_cond.mode == "SPECIFIC":
     --specific '$mode_cond.specific'
@@ -51,15 +59,12 @@
     <conditional name="reference_genome">
         <param name="genome_type_select" type="select" label="Select a source for fasta sequences to be searched for STRs" help="Options are to choose a built-in genome, or choose any history fasta file">
             <option value="indexed">Use a Galaxy server built-in reference genome fasta</option>
-            <option selected="True" value="history">Use any fasta file from the current history</option>
+            <option value="history" selected="True">Use any fasta file from the current history</option>
         </param>
         <when value="indexed">
-            <param name="fasta" type="select" optional="false" multiple="false" label="Choose a built-in, or custom reference genome" 
-                help="If the genome you need is not on the list, add a custom genome or choose a genome fasta from the current history">
-                <options from_data_table="all_fasta">
-                    <filter column="2" type="sort_by"/>
-                    <validator message="No genomes are available" type="no_options"/>
-                </options>
+            <param name="fasta" type="select" multiple="false" label="Choose a built-in genome" 
+                help="If the genome you need is not on the list, upload it and select it as a current history fasta">
+                <options from_data_table="all_fasta"/>
             </param>
         </when>
         <when value="history">
@@ -98,17 +103,17 @@
     </when>
     <when value="NATIVE">
       <param name="outformat" type="select" optional="false" label="Select the required output format" 
-        help="Pytrf can create GFF, CSV or TSV output files">
-        <option value="gff" selected="true">GFF</option>
+        help="Pytrf can create GFF, CSV or TSV output files. Documentation is linked in the help section below">
+        <option value="gff" >GFF</option>
         <option value="csv">Comma separated values</option>
-        <option value="tsv">Tab separated values</option>
+        <option value="tsv" selected="true">Tab separated values</option>
       </param>
+      <param name="monomin" type="integer" value="2" min="2" label="Minimum number of repeats for monomers"/>
       <param name="dimin" type="integer" value="2" min="1" label="Minimum number of repeats for dimers"/>
       <param name="trimin" type="integer" value="2" min="2" label="Minimum number of repeats for trimers"/>
       <param name="tetramin" type="integer" value="2" min="2" label="Minimum number of repeats for tetramers"/>
       <param name="pentamin" type="integer" value="2" min="2" label="Minimum number of repeats for pentamers"/>
       <param name="hexamin" type="integer" value="2" min="2" label="Minimum number of repeats for hexamers"/>
-      <param name="monomin" type="integer" value="2" min="2" label="Minimum number of repeats for monomers"/>
     </when>
     </conditional>
   </inputs>
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/builtinnativetsv_sample	Sun Jul 14 23:52:21 2024 +0000
@@ -0,0 +1,978 @@
+hpat1	507	526	TTTA	4	5	20
+hpat1	1963	1974	CCCCAC	6	2	12
+hpat1	2146	2163	AAAAAC	6	3	18
+hpat1	3236	3247	TTTTTT	6	2	12
+hpat1	4117	4128	GGCAGG	6	2	12
+hpat1	5472	5483	TTTTTG	6	2	12
+hpat1	5584	5595	AGCCCC	6	2	12
+hpat1	7239	7250	GAAAAA	6	2	12
+hpat1	8173	8184	AAAAAA	6	2	12
+hpat1	8275	8302	A	1	28	28
+hpat1	8303	8314	GAGAGA	6	2	12
+hpat1	8744	8755	GGAGGT	6	2	12
+hpat1	8828	8845	AGAAAA	6	3	18
+hpat1	9495	9506	GGAGGT	6	2	12
+hpat1	9571	9595	AAAAC	5	5	25
+hpat1	11623	11634	AGGGAA	6	2	12
+hpat1	11651	11662	AACAAC	6	2	12
+hpat1	12486	12497	CCTCTG	6	2	12
+hpat1	13457	13468	ACCCTA	6	2	12
+hpat1	13471	13488	CCCTAA	6	3	18
+hpat1	13489	13500	CCCCAA	6	2	12
+hpat1	13501	13518	CCCCAG	6	3	18
+hpat1	13523	13546	AACCCT	6	4	24
+hpat1	13559	13582	CCCTAA	6	4	24
+hpat1	13598	13609	ACCCTA	6	2	12
+hpat1	13612	13635	ACCCTA	6	4	24
+hpat1	13641	13658	AACCCC	6	3	18
+hpat1	13659	13670	AACCCT	6	2	12
+hpat1	13691	13708	ACCCTC	6	3	18
+hpat1	13721	13744	AACCCC	6	4	24
+hpat1	13747	13758	CCCAAC	6	2	12
+hpat1	13767	13778	CTCACC	6	2	12
+hpat1	14149	14164	GCCC	4	4	16
+hpat1	14307	14318	GGCGCA	6	2	12
+hpat1	14336	14347	GGCGCA	6	2	12
+hpat1	14384	14395	GGCGCA	6	2	12
+hpat1	14472	14483	GGGGGG	6	2	12
+hpat1	16148	16159	GAGTGG	6	2	12
+hpat1	16547	16558	CTGAGG	6	2	12
+hpat1	17030	17041	CAGGCA	6	2	12
+hpat1	18072	18083	GTCTGG	6	2	12
+hpat1	18120	18131	TCGTCC	6	2	12
+hpat1	18795	18806	GCAGGC	6	2	12
+hpat1	19207	19218	TGCTCC	6	2	12
+hpat1	20148	20159	GGTGGT	6	2	12
+hpat1	22835	22846	GGGAGG	6	2	12
+hpat1	22939	22950	AAGCCA	6	2	12
+hpat1	24658	24669	AAAAAA	6	2	12
+hpat1	25690	25701	ATTCAC	6	2	12
+hpat1	26499	26510	CCCCTG	6	2	12
+hpat1	29330	29341	CCTCCA	6	2	12
+hpat1	30089	30100	CCAGGA	6	2	12
+hpat1	31904	31915	GTGTGT	6	2	12
+hpat1	32448	32459	GGAGGT	6	2	12
+hpat1	32998	33009	TTTTCT	6	2	12
+hpat1	35194	35205	GCCCAG	6	2	12
+hpat1	35991	36002	CTCTCC	6	2	12
+hpat1	36299	36310	TCTCTG	6	2	12
+hpat1	36311	36338	TC	2	14	28
+hpat1	36363	36374	TCTCTC	6	2	12
+hpat1	37142	37153	AAAAAA	6	2	12
+hpat1	37229	37240	ACACAC	6	2	12
+hpat1	38874	38885	AAAAAA	6	2	12
+hpat1	39406	39417	CATGTG	6	2	12
+hpat1	39489	39500	TTCTAG	6	2	12
+hpat1	39620	39642	T	1	23	23
+hpat1	40503	40514	GTACCA	6	2	12
+hpat1	43740	43764	TCCAT	5	5	25
+hpat1	43807	43826	CACTC	5	4	20
+hpat1	43965	43984	TCCAC	5	4	20
+hpat1	44879	44903	TTCCA	5	5	25
+hpat1	45344	45368	TCCAT	5	5	25
+hpat1	45526	45545	CATTC	5	4	20
+hpat1	45554	45573	TCCAT	5	4	20
+hpat1	46948	46967	TCCAT	5	4	20
+hpat1	47107	47131	TTCCA	5	5	25
+hpat1	47141	47160	ATTCC	5	4	20
+hpat1	47303	47322	TCCAC	5	4	20
+hpat1	48673	48692	TCCAT	5	4	20
+hpat1	48806	48825	ATTCC	5	4	20
+hpat1	49681	49700	ATTCC	5	4	20
+hpat1	49856	49875	ATTCC	5	4	20
+hpat1	50315	50334	TCCAC	5	4	20
+hpat1	50775	50794	TCCAT	5	4	20
+hpat1	51153	51172	ATTCC	5	4	20
+hpat1	52508	52519	GCCTTT	6	2	12
+hpat1	53676	53687	TTTCAG	6	2	12
+hpat1	55005	55016	AAAAAC	6	2	12
+hpat1	58746	58757	GCTTAT	6	2	12
+hpat1	59055	59066	TAGAAA	6	2	12
+hpat1	61658	61669	GGAGGT	6	2	12
+hpat1	61733	61747	AAA	3	5	15
+hpat1	62761	62775	AAA	3	5	15
+hpat1	63859	63874	AAAG	4	4	16
+hpat1	65522	65533	TTTTTT	6	2	12
+hpat1	67692	67703	CTCAGA	6	2	12
+hpat1	68526	68537	TGCCTC	6	2	12
+hpat1	68608	68622	TTT	3	5	15
+hpat1	68851	68862	CAGCCT	6	2	12
+hpat1	69176	69187	CCTTAT	6	2	12
+hpat1	72054	72065	CTCTCC	6	2	12
+hpat1	73588	73599	CAAATA	6	2	12
+hpat1	73891	73924	AC	2	17	34
+hpat1	74873	74884	TCTCTT	6	2	12
+hpat1	76181	76192	TTTGTT	6	2	12
+hpat1	77091	77106	ATAC	4	4	16
+hpat1	77427	77441	TTT	3	5	15
+hpat1	78001	78015	TTT	3	5	15
+hpat1	80746	80757	TTTTCA	6	2	12
+hpat1	80794	80805	CTCTCT	6	2	12
+hpat1	81769	81780	CAAAAA	6	2	12
+hpat1	82355	82369	TTT	3	5	15
+hpat1	83934	83945	TAAATG	6	2	12
+hpat1	84620	84631	TCAAAA	6	2	12
+hpat1	86960	86975	TATA	4	4	16
+hpat1	87245	87256	TTTCTG	6	2	12
+hpat1	87483	87494	TTTTCC	6	2	12
+hpat1	87622	87641	TC	2	10	20
+hpat1	87642	87653	TGTGTG	6	2	12
+hpat1	87657	87684	GT	2	14	28
+hpat1	90387	90401	AAA	3	5	15
+hpat1	91598	91609	CTAGAA	6	2	12
+hpat1	92664	92675	AAGAAA	6	2	12
+hpat1	92820	92831	TCAATG	6	2	12
+hpat1	93618	93629	AACAAA	6	2	12
+hpat1	95760	95771	ATCAGT	6	2	12
+hpat1	97920	97931	GTTTTT	6	2	12
+hpat1	98028	98052	T	1	25	25
+hpat1	98115	98126	ACCTCC	6	2	12
+hpat1	100821	100832	CCTGCC	6	2	12
+hpat1	100974	100985	CCCCAC	6	2	12
+hpat1	102825	102836	GGAGTT	6	2	12
+hpat1	103863	103874	CAGAAA	6	2	12
+hpat1	104935	104946	CTATCT	6	2	12
+hpat1	105272	105283	TTCTTT	6	2	12
+hpat1	105982	105993	AAAAAA	6	2	12
+hpat1	106336	106347	CATTAA	6	2	12
+hpat1	108698	108709	AACACA	6	2	12
+hpat1	109407	109421	TTT	3	5	15
+hpat1	109461	109472	AAAACA	6	2	12
+hpat1	109842	109853	AAACAA	6	2	12
+hpat1	109891	109902	TTAAAT	6	2	12
+hpat1	110331	110342	CAGCAA	6	2	12
+hpat1	110481	110492	AAGAAG	6	2	12
+hpat1	111086	111100	TTT	3	5	15
+hpat1	111899	111910	CATGTG	6	2	12
+hpat1	111999	112010	TTTTTT	6	2	12
+hpat1	112079	112090	ACCTCC	6	2	12
+hpat1	112760	112771	AGAAAG	6	2	12
+hpat1	113453	113464	GAAAAA	6	2	12
+hpat1	113543	113563	A	1	21	21
+hpat1	113689	113700	ATGTGT	6	2	12
+hpat1	113844	113855	TTTGTT	6	2	12
+hpat1	119637	119648	TGGTAC	6	2	12
+hpat1	120465	120476	CTAGAA	6	2	12
+hpat1	120548	120559	CACATG	6	2	12
+hpat1	121012	121035	AAAT	4	6	24
+hpat1	124025	124048	TTGT	4	6	24
+hpat1	124049	124066	TGT	3	6	18
+hpat1	124223	124234	TTTTTA	6	2	12
+hpat1	124605	124616	GAAAAA	6	2	12
+hpat1	125716	125727	AAAATT	6	2	12
+hpat1	125753	125764	CTAAAA	6	2	12
+hpat1	125768	125779	CAAAAA	6	2	12
+hpat1	125885	125896	TTTTTT	6	2	12
+hpat1	125906	125923	TTT	3	6	18
+hpat1	126103	126114	ACATAC	6	2	12
+hpat1	127210	127221	TATATA	6	2	12
+hpat1	128149	128164	TTTG	4	4	16
+hpat1	128765	128776	GTATTT	6	2	12
+hpat1	128861	128872	ATTTTC	6	2	12
+hpat1	129323	129334	ACATAT	6	2	12
+hpat1	130114	130125	GTGTTT	6	2	12
+hpat1	131739	131750	TTTGTC	6	2	12
+hpat1	131826	131837	TTCTGA	6	2	12
+hpat1	132617	132628	AGGAAC	6	2	12
+hpat1	134376	134387	AGATTT	6	2	12
+hpat1	135238	135249	ACATAG	6	2	12
+hpat1	135419	135430	CAACTT	6	2	12
+hpat1	136414	136429	CACA	4	4	16
+hpat1	136957	136968	AAAAAG	6	2	12
+hpat1	137030	137041	AAAAAG	6	2	12
+hpat1	137212	137223	GGAGAG	6	2	12
+hpat1	137369	137380	AAACTA	6	2	12
+hpat1	137535	137546	CAAATC	6	2	12
+hpat1	137558	137575	AAA	3	6	18
+hpat1	137727	137738	GAAGTC	6	2	12
+hpat1	138896	138907	AAAAAA	6	2	12
+hpat1	139228	139242	AAA	3	5	15
+hpat1	140264	140275	TGTGTG	6	2	12
+hpat1	140827	140838	TTTTAT	6	2	12
+hpat1	141801	141812	TAATAT	6	2	12
+hpat1	141988	141999	TGTGTG	6	2	12
+hpat1	142002	142017	TATA	4	4	16
+hpat1	143051	143072	TG	2	11	22
+hpat1	144388	144399	AAAAAA	6	2	12
+hpat1	144871	144888	TTG	3	6	18
+hpat1	147608	147619	CTAGAA	6	2	12
+hpat1	148148	148159	AAAAAA	6	2	12
+hpat1	148417	148428	CTCCCT	6	2	12
+hpat1	148859	148870	TAATAA	6	2	12
+hpat1	150186	150217	AC	2	16	32
+hpat1	152016	152027	AAAAAG	6	2	12
+hpat1	152576	152587	TGTGTG	6	2	12
+hpat1	155739	155750	AAGGCA	6	2	12
+hpat1	156308	156325	TTTTAG	6	3	18
+hpat1	157058	157073	TTTA	4	4	16
+hpat1	158960	158971	ACCTAA	6	2	12
+hpat1	159221	159232	TCCCAA	6	2	12
+hpat1	160996	161007	GCTGCC	6	2	12
+hpat1	166906	166925	AATAA	5	4	20
+hpat1	168215	168240	AC	2	13	26
+hpat1	168672	168683	ATTAAC	6	2	12
+hpat1	168738	168767	GAAAA	5	6	30
+hpat1	170405	170416	AAAAAC	6	2	12
+hpat1	171678	171689	AACAAA	6	2	12
+hpat1	171976	171987	AAATCA	6	2	12
+hpat1	173024	173035	ATGGTG	6	2	12
+hpat1	175792	175803	TTCTTT	6	2	12
+hpat1	175804	175818	TTT	3	5	15
+hpat1	176854	176868	TTT	3	5	15
+hpat1	177755	177766	CTGCTT	6	2	12
+hpat1	180030	180047	TTTTTG	6	3	18
+hpat1	180499	180510	CCCACT	6	2	12
+hpat1	180843	180854	ATTTGT	6	2	12
+hpat1	180871	180902	GT	2	16	32
+hpat1	183233	183244	AAATAA	6	2	12
+hpat1	184102	184116	AAT	3	5	15
+hpat1	184259	184282	T	1	24	24
+hpat1	184344	184355	ACCTCC	6	2	12
+hpat1	184870	184885	TCTC	4	4	16
+hpat1	185111	185122	CCCAGG	6	2	12
+hpat1	186379	186390	AGTGGG	6	2	12
+hpat1	186480	186491	ATTCTT	6	2	12
+hpat1	187054	187075	A	1	22	22
+hpat1	188601	188612	TTCTTT	6	2	12
+hpat1	191292	191303	AAAAAA	6	2	12
+hpat1	192318	192329	AAAAAA	6	2	12
+hpat1	192448	192463	TAAA	4	4	16
+hpat1	195173	195184	TGAAAG	6	2	12
+hpat1	198108	198131	T	1	24	24
+hpat1	200074	200088	TTT	3	5	15
+hpat1	202803	202814	ATATGT	6	2	12
+hpat1	202815	202826	ATATAT	6	2	12
+hpat1	202828	202863	TG	2	18	36
+hpat1	202864	202883	TA	2	10	20
+hpat1	203432	203447	TTTG	4	4	16
+hpat1	203519	203530	ACCTCC	6	2	12
+hpat1	206032	206043	ACAAAT	6	2	12
+hpat1	206400	206419	AGGG	4	5	20
+hpat1	206623	206666	GGAA	4	11	44
+hpat1	211255	211266	CAGAAA	6	2	12
+hpat1	213725	213736	AGATAA	6	2	12
+hpat1	217138	217185	AAAT	4	12	48
+hpat1	218421	218432	AAAAAT	6	2	12
+hpat1	218898	218913	TAAA	4	4	16
+hpat1	219257	219274	TTT	3	6	18
+hpat1	222009	222020	AAAAAA	6	2	12
+hpat1	222072	222093	T	1	22	22
+hpat1	222157	222168	GCCTCC	6	2	12
+hpat1	223960	223971	AAAAAA	6	2	12
+hpat1	224548	224559	AACAAC	6	2	12
+hpat1	225191	225208	AAA	3	6	18
+hpat1	225402	225413	AATTAG	6	2	12
+hpat1	225596	225613	AAA	3	6	18
+hpat1	225617	225628	TGACTC	6	2	12
+hpat1	226078	226089	CTGATT	6	2	12
+hpat1	228565	228576	ATAAAA	6	2	12
+hpat1	228923	228940	AAA	3	6	18
+hpat1	230562	230591	TA	2	15	30
+hpat1	230592	230619	GA	2	14	28
+hpat1	232105	232124	CA	2	10	20
+hpat1	232606	232629	AAAAAG	6	4	24
+hpat1	232635	232682	AAGG	4	12	48
+hpat1	233855	233869	AAA	3	5	15
+hpat1	234129	234140	GGAGGT	6	2	12
+hpat1	234204	234215	AAAAAA	6	2	12
+hpat1	234534	234545	AAAAAA	6	2	12
+hpat1	235842	235853	GGAGGT	6	2	12
+hpat1	235922	235937	AAAT	4	4	16
+hpat1	236235	236246	AAACAC	6	2	12
+hpat1	236491	236502	TTCTCT	6	2	12
+hpat1	237108	237147	AC	2	20	40
+hpat1	237543	237558	AAAT	4	4	16
+hpat1	237921	237932	GGAGGT	6	2	12
+hpat1	237998	238009	AAAAAA	6	2	12
+hpat1	239085	239099	TTT	3	5	15
+hpat1	240473	240487	AAA	3	5	15
+hpat1	240934	240959	AT	2	13	26
+hpat1	241171	241182	ATATAT	6	2	12
+hpat1	241668	241687	TTTTG	5	4	20
+hpat1	242103	242118	AGAT	4	4	16
+hpat1	242122	242161	AGAT	4	10	40
+hpat1	243504	243515	GAGAGA	6	2	12
+hpat1	246294	246329	TTTG	4	9	36
+hpat1	246701	246712	TCCTCA	6	2	12
+hpat1	247307	247324	AAA	3	6	18
+hpat1	247883	247894	ATGTTC	6	2	12
+hpat1	248587	248610	GT	2	12	24
+hpat1	248613	248628	GTGT	4	4	16
+hpat1	248666	248677	TGGCAG	6	2	12
+hpat1	250106	250117	CTCAAT	6	2	12
+hpat1	250129	250152	T	1	24	24
+hpat1	250499	250510	AAGAAG	6	2	12
+hpat1	250826	250837	CTCTAC	6	2	12
+hpat1	250951	250970	TAAA	4	5	20
+hpat1	252089	252100	CTCTCT	6	2	12
+hpat1	252706	252717	ATAAAA	6	2	12
+hpat1	252729	252740	AATAAT	6	2	12
+hpat1	253241	253252	CATGTG	6	2	12
+hpat1	254795	254809	TTT	3	5	15
+hpat1	256098	256113	GGAG	4	4	16
+hpat1	256152	256171	AGGG	4	5	20
+hpat1	256552	256587	AC	2	18	36
+hpat1	257215	257226	AAAAAA	6	2	12
+hpat1	257470	257489	TTTTG	5	4	20
+hpat1	259012	259047	AGGG	4	9	36
+hpat1	259068	259079	AAAGAA	6	2	12
+hpat1	259101	259116	AGAA	4	4	16
+hpat1	262098	262109	GGGGTG	6	2	12
+hpat1	262176	262187	TTCTTC	6	2	12
+hpat1	262313	262324	AGGGGC	6	2	12
+hpat1	264201	264224	GGAG	4	6	24
+hpat1	264325	264348	GGAG	4	6	24
+hpat1	264477	264496	GGAG	4	5	20
+hpat1	264513	264536	GGAG	4	6	24
+hpat1	264545	264564	GGAG	4	5	20
+hpat1	264643	264658	GGGA	4	4	16
+hpat1	264660	264675	GGAG	4	4	16
+hpat1	264700	264715	GGAG	4	4	16
+hpat1	265005	265016	GCCCCG	6	2	12
+hpat1	265567	265584	AAT	3	6	18
+hpat1	265803	265817	AAC	3	5	15
+hpat1	265866	265877	TAACAA	6	2	12
+hpat1	265931	265945	AAC	3	5	15
+hpat1	265994	266005	TAACAA	6	2	12
+hpat1	266186	266197	CTGGAA	6	2	12
+hpat1	266759	266778	TAAA	4	5	20
+hpat1	268395	268406	GGAGGT	6	2	12
+hpat1	268907	268918	TTTTTT	6	2	12
+hpat1	269912	269923	ATGGAC	6	2	12
+hpat1	271171	271182	GATTTA	6	2	12
+hpat1	271562	271573	TGAAAA	6	2	12
+hpat1	272103	272114	AGAGAG	6	2	12
+hpat1	272180	272191	AGGAAA	6	2	12
+hpat1	272192	272209	AGAAAA	6	3	18
+hpat1	273132	273143	TTACTG	6	2	12
+hpat1	273507	273536	A	1	30	30
+hpat1	273765	273794	TTTCTT	6	5	30
+hpat1	273795	273815	T	1	21	21
+hpat1	274215	274226	GAGCCT	6	2	12
+hpat1	274597	274608	GCCTGA	6	2	12
+hpat1	274850	274861	TGTTCT	6	2	12
+hpat1	276183	276205	T	1	23	23
+hpat1	277679	277690	ACACAC	6	2	12
+hpat1	278427	278438	AAAAAA	6	2	12
+hpat1	278575	278586	CCTAGT	6	2	12
+hpat1	279697	279708	TTTTTG	6	2	12
+hpat1	282294	282309	AAAG	4	4	16
+hpat1	282512	282523	AAAAAT	6	2	12
+hpat1	282752	282763	TTGCTT	6	2	12
+hpat1	283102	283116	AAA	3	5	15
+hpat1	283586	283597	GGCGGC	6	2	12
+hpat1	283666	283677	CGGCGG	6	2	12
+hpat1	286603	286614	TGACTC	6	2	12
+hpat1	287369	287380	ACCTCC	6	2	12
+hpat1	288276	288287	AGAAAC	6	2	12
+hpat1	290125	290139	AAA	3	5	15
+hpat1	291615	291630	TAAA	4	4	16
+hpat1	291852	291863	TTTTTT	6	2	12
+hpat1	292027	292038	TTTTTT	6	2	12
+hpat1	292194	292209	TTTC	4	4	16
+hpat1	292210	292227	TTT	3	6	18
+hpat1	292583	292594	TTTTCT	6	2	12
+hpat1	292663	292678	TTCC	4	4	16
+hpat1	292687	292698	TCTTTC	6	2	12
+hpat1	293586	293597	TTGAGG	6	2	12
+hpat1	293663	293682	AAAAT	5	4	20
+hpat1	294172	294183	TCTGCC	6	2	12
+hpat1	296514	296531	TTT	3	6	18
+hpat1	297142	297162	ATT	3	7	21
+hpat1	298177	298196	CCTC	4	5	20
+hpat1	298501	298512	TAATTT	6	2	12
+hpat1	298885	298896	CATCCA	6	2	12
+hpat1	299051	299062	TTTTTC	6	2	12
+hpat1	299063	299077	TTT	3	5	15
+hpat1	300065	300076	CCTCAA	6	2	12
+hpat1	300704	300721	TTT	3	6	18
+hpat1	300989	301004	TTTG	4	4	16
+hpat1	301005	301016	TTTTTG	6	2	12
+hpat1	301468	301479	CCACAC	6	2	12
+hpat1	301529	301558	TAT	3	10	30
+hpat1	302529	302540	TTTTTT	6	2	12
+hpat1	303598	303609	AAAAAA	6	2	12
+hpat1	303764	303775	AAAAAT	6	2	12
+hpat1	304750	304761	AAAAAA	6	2	12
+hpat1	305047	305058	AAAAAA	6	2	12
+hpat1	305434	305467	CA	2	17	34
+hpat1	306054	306065	AAAAAA	6	2	12
+hpat1	306497	306508	TGTTGG	6	2	12
+hpat1	307067	307078	AAAAAA	6	2	12
+hpat1	307299	307310	CAGCTC	6	2	12
+hpat1	308588	308607	TATT	4	5	20
+hpat1	309049	309060	AGCAGC	6	2	12
+hpat1	309969	309980	CCACCC	6	2	12
+hpat1	311300	311311	GAGGGG	6	2	12
+hpat1	311654	311665	CCCCCA	6	2	12
+hpat1	312004	312015	GCGGCT	6	2	12
+hpat1	312356	312373	GCC	3	6	18
+hpat1	312410	312421	GAGGAG	6	2	12
+hpat1	314011	314022	CCCTGT	6	2	12
+hpat1	314319	314336	TTT	3	6	18
+hpat1	314344	314355	AACAAC	6	2	12
+hpat1	315088	315099	ATAATA	6	2	12
+hpat1	315897	315908	CAGGCC	6	2	12
+hpat1	317487	317498	AAAAAA	6	2	12
+hpat1	319116	319127	TCTCTC	6	2	12
+hpat1	321259	321276	TTTTGT	6	3	18
+hpat1	321968	321985	TTC	3	6	18
+hpat1	321986	322033	CTT	3	16	48
+hpat1	322044	322055	TTCTTC	6	2	12
+hpat1	322062	322073	TTCTTC	6	2	12
+hpat1	322074	322085	CTCTTC	6	2	12
+hpat1	322086	322100	TTC	3	5	15
+hpat1	322102	322122	TCT	3	7	21
+hpat1	322123	322140	CCT	3	6	18
+hpat1	322154	322168	TTC	3	5	15
+hpat1	323618	323641	TC	2	12	24
+hpat1	323657	323672	AGAG	4	4	16
+hpat1	323766	323777	ACCATC	6	2	12
+hpat1	323846	323857	TCCCCC	6	2	12
+hpat1	324349	324386	TG	2	19	38
+hpat1	324773	324784	TTGTTG	6	2	12
+hpat1	325310	325321	GGAGGT	6	2	12
+hpat1	325761	325772	TTTTTT	6	2	12
+hpat1	325922	325936	TTT	3	5	15
+hpat1	325962	325977	TTTC	4	4	16
+hpat1	325995	326018	TTCT	4	6	24
+hpat1	326132	326143	GCCTCA	6	2	12
+hpat1	326428	326439	ATAACA	6	2	12
+hpat1	326882	326903	CA	2	11	22
+hpat1	328252	328263	ATATGA	6	2	12
+hpat1	328370	328401	T	1	32	32
+hpat1	329377	329396	TC	2	10	20
+hpat1	329866	329883	AAA	3	6	18
+hpat1	330393	330404	CTTCCC	6	2	12
+hpat1	331403	331414	CACACA	6	2	12
+hpat1	331418	331445	AC	2	14	28
+hpat1	331912	331923	AAAAAT	6	2	12
+hpat1	331929	331940	AAAAAG	6	2	12
+hpat1	332957	332968	GCCCCA	6	2	12
+hpat1	333162	333173	CACCCC	6	2	12
+hpat1	334123	334134	TTGTTG	6	2	12
+hpat1	334268	334329	GT	2	31	62
+hpat1	334655	334666	AAAAAA	6	2	12
+hpat1	335468	335479	CTGCAG	6	2	12
+hpat1	335608	335619	GGCCCT	6	2	12
+hpat1	335804	335815	TTGTAT	6	2	12
+hpat1	336042	336053	GAGGCT	6	2	12
+hpat1	336279	336299	A	1	21	21
+hpat1	336448	336474	A	1	27	27
+hpat1	337163	337174	TAAAAA	6	2	12
+hpat1	339091	339102	GGTGAG	6	2	12
+hpat1	340801	340812	TTTTTT	6	2	12
+hpat1	341153	341170	TTTTTA	6	3	18
+hpat1	342133	342147	AAA	3	5	15
+hpat1	343753	343764	GGAGGT	6	2	12
+hpat1	344175	344186	AAAATT	6	2	12
+hpat1	344342	344366	TTTTC	5	5	25
+hpat1	344367	344384	TTT	3	6	18
+hpat1	344599	344610	GGCCTC	6	2	12
+hpat1	345784	345803	A	1	20	20
+hpat1	347339	347350	ACACAG	6	2	12
+hpat1	347894	347905	GGAGGT	6	2	12
+hpat1	347972	347986	AAA	3	5	15
+hpat1	348658	348673	AATG	4	4	16
+hpat1	348739	348766	ATTT	4	7	28
+hpat1	348822	348833	ACCTCT	6	2	12
+hpat1	349404	349415	CCCCTG	6	2	12
+hpat1	350549	350563	AAA	3	5	15
+hpat1	351013	351024	CACATG	6	2	12
+hpat1	351586	351606	T	1	21	21
+hpat1	352174	352193	AAAG	4	5	20
+hpat1	353018	353033	AAAC	4	4	16
+hpat1	353654	353665	TCTCTC	6	2	12
+hpat1	354071	354088	TTT	3	6	18
+hpat1	355441	355464	A	1	24	24
+hpat1	356742	356753	AGATGG	6	2	12
+hpat1	357237	357248	AATTCA	6	2	12
+hpat1	357793	357804	CCACCC	6	2	12
+hpat1	358582	358593	GTGAGA	6	2	12
+hpat1	359129	359148	AAAAG	5	4	20
+hpat1	359151	359166	GGAA	4	4	16
+hpat1	359807	359826	TTAT	4	5	20
+hpat1	360492	360506	AAC	3	5	15
+hpat1	360640	360651	AGAGCA	6	2	12
+hpat1	360676	360687	AACAAC	6	2	12
+hpat1	360993	361004	AAAGAA	6	2	12
+hpat1	361816	361831	AAAC	4	4	16
+hpat1	362642	362653	CTCTTC	6	2	12
+hpat1	362938	362952	TTT	3	5	15
+hpat1	363151	363162	GATCTT	6	2	12
+hpat1	363465	363479	TTT	3	5	15
+hpat1	364493	364507	TTT	3	5	15
+hpat1	365380	365391	GGAGGT	6	2	12
+hpat1	366470	366484	AAA	3	5	15
+hpat1	367830	367841	GGGGTG	6	2	12
+hpat1	367885	367914	TG	2	15	30
+hpat1	367918	367929	GCGCGC	6	2	12
+hpat1	369146	369157	ACACAC	6	2	12
+hpat1	369223	369234	TTGTTT	6	2	12
+hpat1	370082	370093	ATTAGG	6	2	12
+hpat1	370892	370903	TTATTT	6	2	12
+hpat1	371203	371217	TTT	3	5	15
+hpat1	371649	371663	TTT	3	5	15
+hpat1	372440	372451	ACCTCC	6	2	12
+hpat1	373342	373356	AAA	3	5	15
+hpat1	373851	373862	GTTGGA	6	2	12
+hpat1	376041	376052	GGAGGT	6	2	12
+hpat1	376789	376815	A	1	27	27
+hpat1	377249	377260	CCTGAG	6	2	12
+hpat1	377813	377824	AAAAAA	6	2	12
+hpat1	378349	378360	TTTTTT	6	2	12
+hpat1	379129	379140	TTTTTT	6	2	12
+hpat1	379796	379807	TTTTCT	6	2	12
+hpat1	379824	379838	TTT	3	5	15
+hpat1	379902	379913	ACCTCC	6	2	12
+hpat1	380591	380602	AAAAAA	6	2	12
+hpat1	380955	380969	TTT	3	5	15
+hpat1	384331	384342	AAAAAA	6	2	12
+hpat1	384768	384779	TTTTGT	6	2	12
+hpat1	384884	384895	CTTCTG	6	2	12
+hpat1	386748	386759	CTGCAG	6	2	12
+hpat1	387480	387494	AAA	3	5	15
+hpat1	387951	387962	TGGGGG	6	2	12
+hpat1	388197	388208	GGAGGT	6	2	12
+hpat1	388272	388298	A	1	27	27
+hpat1	390061	390072	AATATA	6	2	12
+hpat1	390511	390522	AAAAAA	6	2	12
+hpat1	390939	390974	CA	2	18	36
+hpat1	391213	391224	AACAAC	6	2	12
+hpat1	391730	391741	TTTTTT	6	2	12
+hpat1	392496	392516	A	1	21	21
+hpat1	393018	393029	GGCAGG	6	2	12
+hpat1	393263	393274	TGGAGC	6	2	12
+hpat1	393714	393725	GCTCCT	6	2	12
+hpat1	393907	393918	GCCCCG	6	2	12
+hpat1	394078	394089	CCGCCG	6	2	12
+hpat1	394172	394183	CACTTT	6	2	12
+hpat1	395341	395352	TTTTTT	6	2	12
+hpat1	395519	395530	TGGGAA	6	2	12
+hpat1	396741	396756	TTTA	4	4	16
+hpat1	398200	398211	TTTTTT	6	2	12
+hpat1	399192	399203	AACAAC	6	2	12
+hpat1	399363	399374	TTTCCT	6	2	12
+hpat1	399446	399457	ACTGGC	6	2	12
+hpat1	399792	399806	AAA	3	5	15
+hpat1	399965	399976	AAAAAA	6	2	12
+hpat1	401245	401256	AAAAAC	6	2	12
+hpat1	402032	402049	TTT	3	6	18
+hpat1	403273	403300	TG	2	14	28
+hpat1	405046	405057	GACCCT	6	2	12
+hpat1	406470	406481	CCACCT	6	2	12
+hpat1	406635	406646	CCCCCA	6	2	12
+hpat1	406676	406687	CCCCAT	6	2	12
+hpat1	406729	406740	CCCCCA	6	2	12
+hpat1	406946	406957	AGCAGC	6	2	12
+hpat1	408399	408410	TTTTTT	6	2	12
+hpat1	409295	409306	CGGAGG	6	2	12
+hpat1	409374	409389	ATAT	4	4	16
+hpat1	411079	411090	TTTTTT	6	2	12
+hpat1	411698	411715	TTT	3	6	18
+hpat1	411778	411789	ACCTCC	6	2	12
+hpat1	412137	412148	ATTTTT	6	2	12
+hpat1	412888	412917	T	1	30	30
+hpat1	413750	413761	CCCAGC	6	2	12
+hpat1	415300	415314	TTT	3	5	15
+hpat1	415761	415772	CAGAGG	6	2	12
+hpat1	416538	416549	CCACAT	6	2	12
+hpat1	418814	418825	CAGGGG	6	2	12
+hpat1	422182	422193	GAAACA	6	2	12
+hpat1	423212	423223	AAATAA	6	2	12
+hpat1	424001	424012	GTCTAG	6	2	12
+hpat1	424336	424347	GCCCAG	6	2	12
+hpat1	424668	424679	CAGCTG	6	2	12
+hpat1	425324	425335	GTCACA	6	2	12
+hpat1	425542	425553	CTACTC	6	2	12
+hpat1	427055	427066	TGAGGC	6	2	12
+hpat1	428931	428942	GAGGCA	6	2	12
+hpat1	429287	429298	ATAAAA	6	2	12
+hpat1	430039	430050	ACACAC	6	2	12
+hpat1	430204	430218	AAA	3	5	15
+hpat1	430892	430903	AAAAAC	6	2	12
+hpat1	433017	433040	T	1	24	24
+hpat1	433197	433214	TTT	3	6	18
+hpat1	433986	433997	CATTGA	6	2	12
+hpat1	434454	434465	GTTGTT	6	2	12
+hpat1	434763	434777	AAA	3	5	15
+hpat1	434915	434932	TTT	3	6	18
+hpat1	435226	435243	TTT	3	6	18
+hpat1	435358	435377	A	1	20	20
+hpat1	436225	436239	TTT	3	5	15
+hpat1	436387	436401	TTT	3	5	15
+hpat1	436833	436848	AAAT	4	4	16
+hpat1	437626	437645	AAAT	4	5	20
+hpat1	438925	438936	AAAAAC	6	2	12
+hpat1	439225	439262	A	1	38	38
+hpat1	439559	439573	AAC	3	5	15
+hpat1	439615	439637	T	1	23	23
+hpat1	439700	439711	ACCTCC	6	2	12
+hpat1	440311	440325	AAA	3	5	15
+hpat1	440725	440748	ATT	3	8	24
+hpat1	440749	440760	TTATTT	6	2	12
+hpat1	442019	442036	AAAATA	6	3	18
+hpat1	442318	442329	CAAAAA	6	2	12
+hpat1	442719	442730	ACCTCC	6	2	12
+hpat1	443008	443022	TTT	3	5	15
+hpat1	443088	443099	ACCTCC	6	2	12
+hpat1	443320	443335	TTTA	4	4	16
+hpat1	444386	444400	TTT	3	5	15
+hpat1	445705	445716	TTATTT	6	2	12
+hpat1	445717	445732	ATTT	4	4	16
+hpat1	446505	446519	AAA	3	5	15
+hpat1	446850	446861	AACAAA	6	2	12
+hpat1	447436	447447	TTATTA	6	2	12
+hpat1	448341	448352	CTGCCC	6	2	12
+hpat1	448561	448572	TCGGCA	6	2	12
+hpat1	449223	449234	ATTGTG	6	2	12
+hpat1	449770	449781	TTTTTT	6	2	12
+hpat1	451023	451034	TTTGTT	6	2	12
+hpat1	451475	451490	TGTA	4	4	16
+hpat1	452085	452108	A	1	24	24
+hpat1	452536	452547	ATCAGC	6	2	12
+hpat1	452702	452727	TA	2	13	26
+hpat1	453077	453091	TTT	3	5	15
+hpat1	453851	453862	AAAAAT	6	2	12
+hpat1	453964	453975	CAGATA	6	2	12
+hpat1	455184	455195	ATTTAA	6	2	12
+hpat1	458294	458305	CTGTTG	6	2	12
+hpat1	459372	459383	AGTCTC	6	2	12
+hpat1	459470	459481	CCTGCC	6	2	12
+hpat1	461444	461455	GGAGGT	6	2	12
+hpat1	461519	461533	AAA	3	5	15
+hpat1	462041	462052	AGACTA	6	2	12
+hpat1	462062	462073	CAAAAG	6	2	12
+hpat1	463723	463734	ACCCAG	6	2	12
+hpat1	463956	463975	AAACA	5	4	20
+hpat1	464056	464067	GCATTT	6	2	12
+hpat1	464824	464843	A	1	20	20
+hpat1	466294	466317	TG	2	12	24
+hpat1	466327	466338	GTGTGT	6	2	12
+hpat1	467570	467581	TTTTTT	6	2	12
+hpat1	468461	468472	ACACAC	6	2	12
+hpat1	469650	469661	TGGGAA	6	2	12
+hpat1	470754	470779	TG	2	13	26
+hpat1	471054	471065	GGAGGC	6	2	12
+hpat1	471128	471165	A	1	38	38
+hpat1	471166	471221	TA	2	28	56
+hpat1	471298	471309	TCTCAC	6	2	12
+hpat1	471602	471613	TTGGAG	6	2	12
+hpat1	472084	472095	TACACA	6	2	12
+hpat1	474126	474140	TTT	3	5	15
+hpat1	476040	476051	TTAAAA	6	2	12
+hpat1	476070	476081	TCAAAA	6	2	12
+hpat1	476120	476131	TTTTTT	6	2	12
+hpat1	476421	476435	TTT	3	5	15
+hpat1	476497	476511	TTT	3	5	15
+hpat1	476790	476804	TTT	3	5	15
+hpat1	476979	476996	AAA	3	6	18
+hpat1	478010	478033	AAAACA	6	4	24
+hpat1	478269	478280	TTTTTT	6	2	12
+hpat1	479363	479374	CCACTG	6	2	12
+hpat1	480111	480131	T	1	21	21
+hpat1	480195	480206	CCTCCG	6	2	12
+hpat1	481379	481390	AAACAG	6	2	12
+hpat1	484422	484443	T	1	22	22
+hpat1	484767	484778	TTTTAC	6	2	12
+hpat1	485404	485415	AACAAA	6	2	12
+hpat1	485646	485657	CATCGG	6	2	12
+hpat1	486045	486062	AAA	3	6	18
+hpat1	486722	486743	T	1	22	22
+hpat1	487157	487168	TTTTTT	6	2	12
+hpat1	489258	489269	TCTCAT	6	2	12
+hpat1	490266	490283	AAAACA	6	3	18
+hpat1	490694	490705	TCAGCT	6	2	12
+hpat1	491439	491462	T	1	24	24
+hpat1	491755	491766	TTTTTT	6	2	12
+hpat1	491956	491977	T	1	22	22
+hpat1	492607	492618	CCTGCC	6	2	12
+hpat1	493273	493284	GGAGGC	6	2	12
+hpat1	493346	493357	CACACA	6	2	12
+hpat1	496259	496273	TTT	3	5	15
+hpat1	497786	497813	AC	2	14	28
+hpat1	498279	498290	TTCTTC	6	2	12
+hpat1	498874	498885	TATTTC	6	2	12
+hpat1	498952	498963	CACACA	6	2	12
+hpat1	499113	499124	AGGTGT	6	2	12
+hpat1	500427	500446	AAAC	4	5	20
+hpat1	500448	500459	AAAAAC	6	2	12
+hpat1	500768	500779	CAGCCT	6	2	12
+hpat1	501705	501716	CACACA	6	2	12
+hpat1	501890	501901	AAATGA	6	2	12
+hpat1	504615	504626	TATTTT	6	2	12
+hpat1	504795	504814	T	1	20	20
+hpat1	505013	505033	T	1	21	21
+hpat1	505628	505639	AAACAA	6	2	12
+hpat1	506288	506299	AAAGAC	6	2	12
+hpat1	507516	507527	AGTAAC	6	2	12
+hpat1	508773	508788	TTTA	4	4	16
+hpat1	509939	509961	A	1	23	23
+hpat1	510502	510519	AAA	3	6	18
+hpat1	510633	510647	AAC	3	5	15
+hpat1	510704	510715	GAGGCT	6	2	12
+hpat1	510735	510746	GGAGGT	6	2	12
+hpat1	510788	510808	A	1	21	21
+hpat1	511425	511436	ATAAAA	6	2	12
+hpat1	511729	511743	AAA	3	5	15
+hpat1	511755	511766	AATAAT	6	2	12
+hpat1	511915	511936	A	1	22	22
+hpat1	513045	513060	GAGA	4	4	16
+hpat1	513078	513121	AG	2	22	44
+hpat1	513354	513365	AAAAAC	6	2	12
+hpat1	514714	514725	TTTTGT	6	2	12
+hpat1	515269	515283	TTT	3	5	15
+hpat1	516547	516561	TTT	3	5	15
+hpat1	517637	517648	CATGTA	6	2	12
+hpat1	518248	518259	TTTTTT	6	2	12
+hpat1	518408	518428	T	1	21	21
+hpat1	518943	518954	AAAAAC	6	2	12
+hpat1	519163	519174	TGTTTT	6	2	12
+hpat1	519687	519702	ATAC	4	4	16
+hpat1	520951	520965	AAA	3	5	15
+hpat1	521038	521049	GGAGGT	6	2	12
+hpat1	521113	521144	A	1	32	32
+hpat1	521818	521835	AAA	3	6	18
+hpat1	522338	522349	GAAAAT	6	2	12
+hpat1	524042	524056	TTT	3	5	15
+hpat1	526369	526380	TTCACA	6	2	12
+hpat1	527826	527837	TTTTTA	6	2	12
+hpat1	528955	528966	CCTGAC	6	2	12
+hpat1	529165	529176	AACTCC	6	2	12
+hpat1	529760	529771	AGCGGC	6	2	12
+hpat1	532566	532585	CT	2	10	20
+hpat1	532765	532779	TAT	3	5	15
+hpat1	534656	534681	T	1	26	26
+hpat1	535074	535085	TTTTAT	6	2	12
+hpat1	535729	535744	GAAA	4	4	16
+hpat1	535746	535773	AAAG	4	7	28
+hpat1	535782	535793	AGAGAG	6	2	12
+hpat1	536150	536161	TTTAGT	6	2	12
+hpat1	538740	538751	AAATAT	6	2	12
+hpat1	538773	538792	AAGG	4	5	20
+hpat1	538800	538811	TTTTTT	6	2	12
+hpat1	539339	539350	GGAGGT	6	2	12
+hpat1	539414	539433	A	1	20	20
+hpat1	540074	540085	AAAAAA	6	2	12
+hpat1	540498	540509	TGTTTT	6	2	12
+hpat1	541889	541900	AACAAC	6	2	12
+hpat1	542948	542959	AAAAAA	6	2	12
+hpat1	542960	542975	ATAT	4	4	16
+hpat1	543137	543148	CAACAA	6	2	12
+hpat1	543959	543970	GGAGGT	6	2	12
+hpat1	544033	544050	AAAAAC	6	3	18
+hpat1	544054	544068	ACC	3	5	15
+hpat1	544522	544533	TTGCTG	6	2	12
+hpat1	544658	544669	TTTTTT	6	2	12
+hpat1	546135	546146	CAAAAA	6	2	12
+hpat1	546382	546393	TTTTTC	6	2	12
+hpat1	546493	546504	AGCCTC	6	2	12
+hpat1	547434	547445	AAATAA	6	2	12
+hpat1	549626	549637	GAGAGG	6	2	12
+hpat1	550639	550658	TGTT	4	5	20
+hpat1	550723	550734	ACCTCC	6	2	12
+hpat1	551639	551650	TTTTAT	6	2	12
+hpat1	552257	552280	AAAAAT	6	4	24
+hpat1	552764	552781	TTA	3	6	18
+hpat1	552848	552859	ACCTCC	6	2	12
+hpat1	556060	556071	GGTTTT	6	2	12
+hpat1	556296	556310	TTT	3	5	15
+hpat1	556975	557006	AAAT	4	8	32
+hpat1	557124	557138	TTT	3	5	15
+hpat1	558010	558021	TGGCTC	6	2	12
+hpat1	558471	558482	GATCTG	6	2	12
+hpat1	562681	562692	CTGTAA	6	2	12
+hpat1	562697	562708	TTTTCT	6	2	12
+hpat1	563925	563936	GAGGCT	6	2	12
+hpat1	564030	564044	AAA	3	5	15
+hpat1	564727	564758	AC	2	16	32
+hpat1	565343	565354	CATGTG	6	2	12
+hpat1	565426	565437	TTCTAG	6	2	12
+hpat1	566572	566583	AGGAGA	6	2	12
+hpat1	566984	566995	AAAACA	6	2	12
+hpat1	568068	568091	T	1	24	24
+hpat1	569331	569342	GGAGGT	6	2	12
+hpat1	569405	569416	AAAAAT	6	2	12
+hpat1	570032	570043	TTTTTT	6	2	12
+hpat1	570541	570558	AAT	3	6	18
+hpat1	571280	571291	TGTATA	6	2	12
+hpat1	571318	571333	TGTG	4	4	16
+hpat1	571613	571633	T	1	21	21
+hpat1	572152	572163	TTATTA	6	2	12
+hpat1	573516	573530	TTG	3	5	15
+hpat1	573851	573862	TTTTTT	6	2	12
+hpat1	575452	575471	AATTA	5	4	20
+hpat1	575472	575483	AATTTT	6	2	12
+hpat1	575669	575680	ATTTTT	6	2	12
+hpat1	575802	575813	ATTTTT	6	2	12
+hpat1	575871	575894	AC	2	12	24
+hpat1	577212	577226	TTT	3	5	15
+hpat1	577377	577388	TTTGTA	6	2	12
+hpat1	578002	578013	CCTTTC	6	2	12
+hpat1	578512	578523	TTTTTT	6	2	12
+hpat1	578671	578702	TA	2	16	32
+hpat1	578703	578717	TTT	3	5	15
+hpat1	578849	578866	TTT	3	6	18
+hpat1	579478	579489	AGAGAG	6	2	12
+hpat1	579493	579504	GGGTCT	6	2	12
+hpat1	580201	580220	A	1	20	20
+hpat1	580329	580340	TTTTTA	6	2	12
+hpat1	580420	580431	ACCTCC	6	2	12
+hpat1	580977	580988	AAAAAA	6	2	12
+hpat1	581225	581242	TTA	3	6	18
+hpat1	581452	581463	CAGGGG	6	2	12
+hpat1	581509	581520	GGGTGG	6	2	12
+hpat1	582118	582141	AAAACA	6	4	24
+hpat1	583541	583556	TGTA	4	4	16
+hpat1	584746	584763	TTT	3	6	18
+hpat1	585485	585496	GAGGCT	6	2	12
+hpat1	587188	587199	GGAGGT	6	2	12
+hpat1	587261	587275	AAA	3	5	15
+hpat1	588890	588901	CAGCCC	6	2	12
+hpat1	589708	589719	CCCCAC	6	2	12
+hpat1	589883	589894	CACACT	6	2	12
+hpat1	591642	591653	CCAGGT	6	2	12
+hpat1	591776	591787	GGGCTT	6	2	12
+hpat1	591897	591908	CAGGGA	6	2	12
+hpat1	592321	592332	GGAGGC	6	2	12
+hpat1	592395	592412	AAA	3	6	18
+hpat1	593794	593805	TCATCT	6	2	12
+hpat1	595349	595360	CGGCGG	6	2	12
+hpat1	595567	595578	TCCCGC	6	2	12
+hpat1	595602	595613	CACCAC	6	2	12
+hpat1	595622	595636	CGC	3	5	15
+hpat1	595738	595749	TCCTCG	6	2	12
+hpat1	597756	597767	TGAGAC	6	2	12
+hpat1	599394	599405	GTGTGC	6	2	12
+hpat1	600006	600025	TC	2	10	20
+hpat1	600026	600037	TATATA	6	2	12
+hpat1	600041	600052	ATATAT	6	2	12
+hpat1	600118	600129	TATATA	6	2	12
+hpat1	601868	601882	TTT	3	5	15
+hpat1	603188	603199	GAAGTG	6	2	12
+hpat1	608484	608499	TGAA	4	4	16
+hpat1	608985	608996	GCTGTG	6	2	12
+hpat1	609780	609795	AAAC	4	4	16
+hpat1	610192	610203	TATCTC	6	2	12
+hpat1	612381	612392	TCACCA	6	2	12
+hpat1	613446	613457	CCATCC	6	2	12
+hpat1	613964	613975	TTCTTG	6	2	12
+hpat1	615611	615622	AGAAGG	6	2	12
+hpat1	615632	615643	AAGAAG	6	2	12
+hpat1	616140	616161	A	1	22	22
+hpat1	617678	617697	TTTTG	5	4	20
+hpat1	620321	620332	GCCAGC	6	2	12
+hpat1	620338	620349	GGGGTG	6	2	12
+hpat1	621050	621061	AGGGCC	6	2	12
+hpat1	621851	621862	TCTCTC	6	2	12
+hpat1	622005	622016	GATTTT	6	2	12
+hpat1	622557	622571	ACC	3	5	15
+hpat1	622586	622606	CAC	3	7	21
+hpat1	623139	623150	ACACAA	6	2	12
+hpat1	623378	623389	TGGACA	6	2	12
+hpat1	623706	623723	AAATAA	6	3	18
+hpat1	624013	624030	AAA	3	6	18
+hpat1	624648	624662	TTT	3	5	15
+hpat1	624724	624735	ACCTCC	6	2	12
+hpat1	624982	624993	TTTTTT	6	2	12
+hpat1	625129	625146	TTTTAT	6	3	18
+hpat1	625595	625606	AGCCAG	6	2	12
+hpat1	627030	627041	GAGGCT	6	2	12
+hpat1	627187	627198	AATAAC	6	2	12
+hpat1	629078	629092	TGA	3	5	15
+hpat1	631394	631405	AACAAC	6	2	12
+hpat1	631429	631443	AAA	3	5	15
+hpat1	632704	632731	A	1	28	28
+hpat1	632732	632743	GTGGGG	6	2	12
+hpat1	633722	633733	ACTCTC	6	2	12
+hpat1	634319	634330	CTTCCT	6	2	12
+hpat1	634612	634623	TCCTTG	6	2	12
+hpat1	635904	635918	AAA	3	5	15
+hpat1	637886	637900	AAA	3	5	15
+hpat1	639391	639402	CGCCGC	6	2	12
+hpat1	639742	639759	GCGGGG	6	3	18
+hpat1	640687	640698	CCTTTC	6	2	12
+hpat1	642772	642783	AAAAAA	6	2	12
+hpat1	646664	646681	AAA	3	6	18
+hpat1	649141	649152	AGCTGG	6	2	12
+hpat1	649225	649236	GTGTGT	6	2	12
+hpat1	649865	649876	GCGAGG	6	2	12
+hpat1	650432	650443	CCGCCC	6	2	12
+hpat1	650444	650455	CGCCGC	6	2	12
+hpat1	651677	651688	CCCCTC	6	2	12
+hpat1	651819	651830	TGCCTC	6	2	12
+hpat1	652399	652410	GATAGT	6	2	12
+hpat1	652602	652613	CTCCTT	6	2	12
+hpat1	653054	653065	GGAAGG	6	2	12
+hpat1	655086	655103	TACTGC	6	3	18
+hpat1	655717	655728	AGCCTC	6	2	12
+hpat1	656009	656020	CCTCCC	6	2	12
+hpat1	657238	657249	AGGCCA	6	2	12
+hpat1	657509	657520	TTGGAC	6	2	12
+hpat1	657610	657621	TGGACC	6	2	12
+hpat1	658606	658633	GT	2	14	28
+hpat1	658934	658949	CCCT	4	4	16
+hpat1	660059	660070	GTGTCA	6	2	12
+hpat1	660332	660343	GCTGTA	6	2	12
+hpat1	660404	660415	CTCTCT	6	2	12
+hpat1	660566	660577	TCTCTC	6	2	12
+hpat1	660592	660621	TC	2	15	30
+hpat1	660622	660647	AC	2	13	26
+hpat1	660896	660927	GT	2	16	32
+hpat1	660985	660996	TGGGGA	6	2	12
+hpat1	661631	661642	CCCCCG	6	2	12
+hpat1	661936	661947	GGCTCG	6	2	12
+hpat1	661953	661964	GGCGGC	6	2	12
+hpat1	662189	662200	CAGAGA	6	2	12
+hpat1	662391	662402	ACACAC	6	2	12
+hpat1	662617	662628	CCTCTC	6	2	12
+hpat1	663604	663618	AAA	3	5	15
+hpat1	663771	663791	AAT	3	7	21
+hpat1	664081	664092	TTTTTT	6	2	12
+hpat1	667668	667679	AATGGG	6	2	12
+hpat1	668025	668036	GTGTAA	6	2	12
+hpat1	668718	668729	CCATTT	6	2	12
+hpat1	669900	669914	AAA	3	5	15
+hpat1	670208	670219	TTTTTA	6	2	12
+hpat1	671191	671202	GAGGGG	6	2	12
+hpat1	671869	671886	TACAAA	6	3	18
+hpat1	671887	671925	TAA	3	13	39
+hpat1	671964	671975	CTGAGG	6	2	12
+hpat1	672093	672104	AACAAC	6	2	12
+hpat1	672105	672116	AAAGAA	6	2	12
+hpat1	672622	672639	AAA	3	6	18
+hpat1	675613	675624	AAGTGT	6	2	12
+hpat1	675731	675742	GGCACA	6	2	12
+hpat1	676339	676350	ATTTCT	6	2	12
+hpat1	676352	676363	TTTATT	6	2	12
+hpat1	677003	677036	GT	2	17	34
+hpat1	677457	677468	AGAAAG	6	2	12
+hpat1	678388	678399	TGAAAC	6	2	12
+hpat1	678858	678869	TTTTTC	6	2	12
+hpat1	682346	682361	TTAT	4	4	16
+hpat1	684578	684589	AATAAC	6	2	12
+hpat1	684953	684964	AGAGCC	6	2	12
+hpat1	685312	685323	GGAGGT	6	2	12
+hpat1	685386	685413	A	1	28	28
+hpat1	686068	686079	CGGGGG	6	2	12
+hpat1	686825	686842	TTT	3	6	18
+hpat1	686899	686910	CAGCCT	6	2	12
+hpat1	687586	687605	TCAA	4	5	20
+hpat1	687651	687662	AGGGAG	6	2	12
+hpat1	688067	688078	TCCTGG	6	2	12
+hpat1	689018	689037	AAAT	4	5	20
+hpat1	690132	690143	TTTCTT	6	2	12
+hpat1	690303	690314	AAGAGG	6	2	12
+hpat1	693707	693718	AGGGGC	6	2	12
+hpat1	694426	694437	TTTTGT	6	2	12
+hpat1	694506	694517	CAGCCT	6	2	12
+hpat1	694598	694609	TTTTTT	6	2	12
+hpat1	694903	694914	TATTTA	6	2	12
+hpat1	695265	695276	CAAGGC	6	2	12
+hpat1	696496	696507	CGCCCC	6	2	12
+hpat1	696718	696729	CGCCCC	6	2	12
+hpat1	697067	697078	TGGGGG	6	2	12
+hpat1	697568	697589	GT	2	11	22
+hpat1	698582	698593	TCTGAT	6	2	12
+hpat1	698951	698962	AATTGA	6	2	12
+hpat1	699157	699171	TTT	3	5	15
--- a/tool-data/all_fasta.loc.sample	Sun Jul 14 03:55:47 2024 +0000
+++ b/tool-data/all_fasta.loc.sample	Sun Jul 14 23:52:21 2024 +0000
@@ -15,4 +15,4 @@
 #Your all_fasta.loc file should contain an entry for each individual
 #fasta file. So there will be multiple fasta files for each build,
 #such as with hg19 above.
-hgtest	hgtest	test	humsamp.fa
+#
--- a/tool_data_table_conf.xml.sample	Sun Jul 14 03:55:47 2024 +0000
+++ b/tool_data_table_conf.xml.sample	Sun Jul 14 23:52:21 2024 +0000
@@ -2,6 +2,6 @@
     <!-- Locations of all fasta files under genome directory -->
     <table name="all_fasta" comment_char="#">
         <columns>value, dbkey, name, path</columns>
-        <file path="test-data/all_fasta.loc" />
+        <file path="tool-data/all_fasta.loc" />
     </table>
 </tables>
--- a/tool_data_table_conf.xml.test	Sun Jul 14 03:55:47 2024 +0000
+++ b/tool_data_table_conf.xml.test	Sun Jul 14 23:52:21 2024 +0000
@@ -1,5 +1,5 @@
 <tables>
-    <!-- Locations of dbkeys and len files under genome directory -->
+    <!-- Locations of all fasta files under genome directory -->
     <table name="all_fasta" comment_char="#">
         <columns>value, dbkey, name, path</columns>
         <file path="${__HERE__}/test-data/all_fasta.loc" />