Repository 'nastiseq'
hg clone https://toolshed.g2.bx.psu.edu/repos/rnateam/nastiseq

Changeset 0:0a0bba8e1823 (2017-02-22)
Commit message:
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/rna_tools/nastiseq commit 8b472e8680bb0ae5d11ee48b642ab305f9333a48
added:
nastiseq.xml
test-data/input_TAIR10_annotation.gtf
test-data/input_positive_pair.tsv
test-data/input_read_count_asmt.tsv
test-data/input_read_count_smt.tsv
test-data/output_neworphan.tsv
test-data/output_newpairs.tsv
test-data/test.r
b
diff -r 000000000000 -r 0a0bba8e1823 nastiseq.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/nastiseq.xml Wed Feb 22 07:32:08 2017 -0500
[
@@ -0,0 +1,175 @@
+<tool id="nastiseq" name="NASTIseq" version="1.0">
+  <description>Identify cis-NATs using ssRNA-seq</description>
+
+  <requirements>
+    <requirement type="package" version="1.0">r-nastiseq</requirement>
+  </requirements>
+  <stdio>
+        <regex match="Execution halted"
+           source="both"
+           level="fatal"
+           description="Execution halted." />
+        <regex match="Error in"
+           source="both"
+           level="fatal"
+           description="An undefined error occured, please check your intput carefully and contact your administrator." />
+  </stdio>
+  <command>
+<![CDATA[
+      Rscript '$script_file'
+]]>
+  </command>
+  <configfiles>
+    <configfile name="script_file">
+          library(NASTIseq)
+
+          genepos = read.delim("${annotation}", header=FALSE, comment.char="#")
+          colnames(genepos) = c("seqname", "source", "feature", "start", "end", "score", "strand", "frame", "attributes")
+          genepos = subset(genepos, feature=="gene")
+
+          get_id = function(attri){
+              gene_info = strsplit(attri, ";")[[1]][1]
+              gene_id = strsplit(gene_info, " ")[[1]][2]
+              gene_id = gsub('\"', '', gene_id)
+              return(gene_id)
+           }
+
+          genepos\$attributes = as.character(lapply(as.character(genepos\$attributes), get_id))
+
+          pospairs = read.table("${positive_pair}", sep = "\t", as.is = TRUE)
+
+          smat = as.matrix(read.table("${count_smt}",  sep = "\t",  row.names = 1))
+
+          asmat = as.matrix(read.table("${count_asmt}",  sep = "\t",  row.names = 1))
+
+          WRscore = getNASTIscore(smat, asmat)
+
+          negpairs = getnegativepairs(genepos)
+
+          WRpred = NASTIpredict(smat,asmat, pospairs, negpairs)
+
+          WRpred_rocr = prediction(WRpred\$predictions,WRpred\$labels)
+
+          thr = defineFDR(WRpred_rocr,0.05)
+
+          WR_names = FindNATs(WRscore, thr, pospairs, genepos)
+
+          write.table(WR_names\$newpairs, file = "output_newpairs.tsv", row.names = FALSE, col.names = FALSE, sep = "\t", quote = FALSE)
+
+          write.table(WR_names\$neworphan, file = "output_neworphan.tsv", row.names = FALSE, col.names = FALSE, sep = "\t", quote = FALSE)
+
+    </configfile>
+  </configfiles>
+  <inputs>
+    <param name="annotation" type="data" format="gtf" label="Annotation file"
+        help="The gene ids should be in agreement in the files of annotation, known pairs and read count.">
+    </param>
+    <param name="positive_pair" type="data" format="tabular" label="Known pairs"
+        help="A known pair of cis-natural antisense transcripts">
+    </param>
+    <param name="count_smt" type="data" format="tabular" label="Read count of sense strand"
+        help="">
+    </param>
+    <param name="count_asmt" type="data" format="tabular" label="Read count of antisense strand"
+        help="">
+    </param>
+  </inputs>
+  <outputs>
+      <data name="newpairs" format="tabular"
+      from_work_dir="output_newpairs.tsv"
+      label="${tool.name} on ${on_string}: New pairs">
+      </data>
+      <data name="neworphan" format="tabular"
+      from_work_dir="output_neworphan.tsv"
+      label="${tool.name} on ${on_string}: New orphans">
+      </data>
+  </outputs>
+  <tests>
+    <test>
+        <param name="annotation" value="input_TAIR10_annotation.gtf" ftype="gtf" />
+        <param name="positive_pair" value="input_positive_pair.tsv" ftype="tabular" />
+        <param name="count_smt" value="input_read_count_smt.tsv" ftype="tabular" />
+        <param name="count_asmt" value="input_read_count_asmt.tsv" ftype="tabular" />
+        <output name="newpairs" file="output_newpairs.tsv" ftype="tabular"/>
+        <output name="neworphan" file="output_neworphan.tsv" ftype="tabular"/>
+    </test>
+  </tests>
+<help>
+<![CDATA[
+.. class:: infomark
+
+**What it does**
+
+  Pairs of RNA molecules transcribed from partially or entirely complementary loci
+  are called cis-natural antisense transcripts (cis-NATs),
+  and they play key roles in the regulation of gene expression in many organisms.
+  A promising experimental tool for profiling sense and antisense transcription
+  is strand-specific RNA sequencing (ssRNA-seq). `NASTIseq`_  is to identify
+  cis-NATs using ssRNA-seq. `NASTIseq`_  is based on model comparison that incorporates
+  the inherent variable efficiency of generating perfectly strand-specific libraries.
+  Applying the method to the ssRNA-seq data from whole root and
+  cell-type specific Arabidopsis libraries confirmed most of
+  the known cis-NAT pairs and identified hundreds of additional cis-NAT pairs.
+
+.. _NASTIseq: https://ohlerlab.mdc-berlin.de/software/NASTIseq_104/
+
+.. class:: infomark
+
+**Inputs**
+
+  ``Annotation file``:  the annotation in `gtf`_ format
+
+  .. _gtf: http://www.ensembl.org/info/website/upload/gff.html
+
+  ``Known pairs``: a table of two  column  matrix,  with  each  row  contains  the
+  names of a known pair of cis-natural antisense transcripts. Example as following::
+
+          AT2G46910       AT2G46915
+          AT3G12250       AT3G12260
+          AT5G50315       AT5G50320
+
+  ``Read count of sense strand``: a table of N by M matrix of read count for reads that mapped
+  to  the  sense  strand.   N  is  the  number  of  gene  loci.   M  is  the
+  number of biological replicates in the sample.  Each
+  rowname must be a unique locus name. Example as following::
+
+          AT1G38440       0       2       0
+          AT1G43171       2       8       1
+          AT1G67670       3       7       0
+
+  ``Read count of antisense strand``: a table of N by M matrix of read count for reads that mapped
+  to  the  antisense  strand.    N  is  the  number  of  gene  loci.   M  is  the
+  number of biological replicates in the sample.  Each
+  rowname must be a unique locus name. Example as following::
+
+          AT1G38440       0       0       0
+          AT1G43171       0       0       0
+          AT1G67670       0       0       2
+
+  Read counts can be obtained using popular software such as `RSamtools`_.
+
+.. _RSamtools: http://bioconductor.org/packages/release/bioc/html/Rsamtools.html
+
+.. class:: infomark
+
+**Outputs**
+
+  ``New pairs``: a table of two  column  matrix,  with  each  row  contains  the
+  names of a new pair of cis-natural antisense transcripts. Example as following::
+
+        AT1G76630       AT1G76640
+        AT2G06045       AT2G06050
+        AT4G30100       AT4G30110
+
+
+  ``New orphans``: a list of new orphan transcripts. Example as following::
+
+        ATMG00030
+        AT5G49440
+        AT2G11240
+]]>
+</help>
+<citations>
+    <citation type="doi">10.1101/gr.149310.112</citation>
+</citations>
+</tool>
b
diff -r 000000000000 -r 0a0bba8e1823 test-data/input_TAIR10_annotation.gtf
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/input_TAIR10_annotation.gtf Wed Feb 22 07:32:08 2017 -0500
b
b'@@ -0,0 +1,33598 @@\n+#!genome-build TAIR10 gene\n+Chr1\tprotein_coding\tgene\t3631\t5899\t.\t+\t.\tgene_id "AT1G01010";\n+Chr1\tprotein_coding\tgene\t5928\t8737\t.\t-\t.\tgene_id "AT1G01020";\n+Chr1\tprotein_coding\tgene\t11649\t13714\t.\t-\t.\tgene_id "AT1G01030";\n+Chr1\tprotein_coding\tgene\t23146\t31227\t.\t+\t.\tgene_id "AT1G01040";\n+Chr1\tmiRNA\tgene\t28500\t28706\t.\t+\t.\tgene_id "AT1G01046";\n+Chr1\tprotein_coding\tgene\t31170\t33153\t.\t-\t.\tgene_id "AT1G01050";\n+Chr1\tprotein_coding\tgene\t33379\t37871\t.\t-\t.\tgene_id "AT1G01060";\n+Chr1\tprotein_coding\tgene\t38752\t40944\t.\t-\t.\tgene_id "AT1G01070";\n+Chr1\tprotein_coding\tgene\t44677\t44787\t.\t+\t.\tgene_id "AT1G01073";\n+Chr1\tprotein_coding\tgene\t45296\t47019\t.\t-\t.\tgene_id "AT1G01080";\n+Chr1\tprotein_coding\tgene\t47485\t49286\t.\t-\t.\tgene_id "AT1G01090";\n+Chr1\tprotein_coding\tgene\t50075\t51199\t.\t-\t.\tgene_id "AT1G01100";\n+Chr1\tprotein_coding\tgene\t52239\t54692\t.\t+\t.\tgene_id "AT1G01110";\n+Chr1\tprotein_coding\tgene\t56624\t56740\t.\t+\t.\tgene_id "AT1G01115";\n+Chr1\tprotein_coding\tgene\t57269\t59167\t.\t-\t.\tgene_id "AT1G01120";\n+Chr1\tprotein_coding\tgene\t61905\t63811\t.\t-\t.\tgene_id "AT1G01130";\n+Chr1\tprotein_coding\tgene\t64166\t67625\t.\t-\t.\tgene_id "AT1G01140";\n+Chr1\tprotein_coding\tgene\t70115\t72138\t.\t-\t.\tgene_id "AT1G01150";\n+Chr1\tprotein_coding\tgene\t72339\t74096\t.\t+\t.\tgene_id "AT1G01160";\n+Chr1\tprotein_coding\tgene\t73931\t74737\t.\t-\t.\tgene_id "AT1G01170";\n+Chr1\tprotein_coding\tgene\t75583\t76758\t.\t+\t.\tgene_id "AT1G01180";\n+Chr1\tmiRNA\tgene\t78932\t79032\t.\t-\t.\tgene_id "AT1G01183";\n+Chr1\tprotein_coding\tgene\t83045\t84864\t.\t-\t.\tgene_id "AT1G01190";\n+Chr1\tprotein_coding\tgene\t86515\t88213\t.\t-\t.\tgene_id "AT1G01200";\n+Chr1\tprotein_coding\tgene\t88898\t89745\t.\t+\t.\tgene_id "AT1G01210";\n+Chr1\tprotein_coding\tgene\t91376\t95651\t.\t+\t.\tgene_id "AT1G01220";\n+Chr1\tprotein_coding\tgene\t95987\t97407\t.\t+\t.\tgene_id "AT1G01225";\n+Chr1\tprotein_coding\tgene\t97456\t99240\t.\t+\t.\tgene_id "AT1G01230";\n+Chr1\tprotein_coding\tgene\t99894\t101834\t.\t+\t.\tgene_id "AT1G01240";\n+Chr1\tprotein_coding\tgene\t104491\t105330\t.\t-\t.\tgene_id "AT1G01250";\n+Chr1\tprotein_coding\tgene\t108946\t111609\t.\t+\t.\tgene_id "AT1G01260";\n+Chr1\ttRNA\tgene\t111890\t111961\t.\t-\t.\tgene_id "AT1G01270";\n+Chr1\tprotein_coding\tgene\t112263\t113947\t.\t+\t.\tgene_id "AT1G01280";\n+Chr1\tprotein_coding\tgene\t114286\t116108\t.\t+\t.\tgene_id "AT1G01290";\n+Chr1\tprotein_coding\tgene\t116943\t118764\t.\t+\t.\tgene_id "AT1G01300";\n+Chr1\tprotein_coding\tgene\t119397\t119997\t.\t+\t.\tgene_id "AT1G01305";\n+Chr1\tprotein_coding\tgene\t120154\t121130\t.\t+\t.\tgene_id "AT1G01310";\n+Chr1\tprotein_coding\tgene\t121067\t130577\t.\t-\t.\tgene_id "AT1G01320";\n+Chr1\tprotein_coding\tgene\t132328\t135831\t.\t-\t.\tgene_id "AT1G01340";\n+Chr1\tprotein_coding\tgene\t136124\t138162\t.\t+\t.\tgene_id "AT1G01350";\n+Chr1\tprotein_coding\tgene\t138513\t139568\t.\t+\t.\tgene_id "AT1G01355";\n+Chr1\tprotein_coding\tgene\t141971\t143183\t.\t+\t.\tgene_id "AT1G01360";\n+Chr1\tprotein_coding\tgene\t143564\t145684\t.\t+\t.\tgene_id "AT1G01370";\n+Chr1\tprotein_coding\tgene\t147153\t147942\t.\t+\t.\tgene_id "AT1G01380";\n+Chr1\tprotein_coding\tgene\t148120\t149806\t.\t-\t.\tgene_id "AT1G01390";\n+Chr1\tprotein_coding\tgene\t150689\t152210\t.\t-\t.\tgene_id "AT1G01400";\n+Chr1\tprotein_coding\tgene\t153113\t154198\t.\t+\t.\tgene_id "AT1G01410";\n+Chr1\tprotein_coding\tgene\t154492\t156011\t.\t-\t.\tgene_id "AT1G01420";\n+Chr1\tprotein_coding\tgene\t156801\t158655\t.\t-\t.\tgene_id "AT1G01430";\n+Chr1\tprotein_coding\tgene\t159856\t162572\t.\t-\t.\tgene_id "AT1G01440";\n+Chr1\tncRNA\tgene\t163419\t166239\t.\t+\t.\tgene_id "AT1G01448";\n+Chr1\tprotein_coding\tgene\t164105\t165517\t.\t-\t.\tgene_id "AT1G01450";\n+Chr1\tprotein_coding\tgene\t166589\t167842\t.\t-\t.\tgene_id "AT1G01453";\n+Chr1\tprotein_coding\tgene\t168723\t171165\t.\t+\t.\tgene_id "AT1G01460";\n+Chr1\tprotein_coding\tgene\t172146\t172948\t.\t-\t.\tgene_id "AT1G01470";\n+Chr1\tprotein_coding\tgene\t173251\t173466\t.\t+\t.\tgene_id "AT1G01471";\n+Chr1\tprotein_coding\tgene\t175782\t178400\t.\t+\t.\tgene_id "AT1G01480";\n+Chr1\tprotein_coding\tgene\t180059\t182358\t.\t-\t.\tgene_id "AT1G01490";\n+Chr1\tprotein_coding\tgene\t185133\t186923\t.\t+\t.\tgene_id "AT1G01500";\n+Chr1\tprotein_coding\tgene\t187211\t190056\t.\t+\t.\tgene_id "AT1G01510";\n+Chr1\tprotein_coding\tgene\t190'..b'ng\tgene\t228573\t229085\t.\t-\t.\tgene_id "ATMG00820";\n+ChrM\tprotein_coding\tgene\t231894\t233042\t.\t-\t.\tgene_id "ATMG00830";\n+ChrM\tprotein_coding\tgene\t234538\t234903\t.\t-\t.\tgene_id "ATMG00840";\n+ChrM\tprotein_coding\tgene\t235338\t235661\t.\t+\t.\tgene_id "ATMG00850";\n+ChrM\tprotein_coding\tgene\t235916\t236392\t.\t+\t.\tgene_id "ATMG00860";\n+ChrM\tprotein_coding\tgene\t236927\t237481\t.\t-\t.\tgene_id "ATMG00870";\n+ChrM\tprotein_coding\tgene\t237053\t237616\t.\t+\t.\tgene_id "ATMG00880";\n+ChrM\tprotein_coding\tgene\t239636\t239956\t.\t-\t.\tgene_id "ATMG00890";\n+ChrM\tprotein_coding\tgene\t239988\t240758\t.\t-\t.\tgene_id "ATMG00900";\n+ChrM\tprotein_coding\tgene\t241306\t241953\t.\t-\t.\tgene_id "ATMG00910";\n+ChrM\tprotein_coding\tgene\t249272\t249919\t.\t-\t.\tgene_id "ATMG00920";\n+ChrM\ttRNA\tgene\t250080\t250153\t.\t-\t.\tgene_id "ATMG00930";\n+ChrM\tprotein_coding\tgene\t251403\t251897\t.\t-\t.\tgene_id "ATMG00940";\n+ChrM\ttRNA\tgene\t254505\t254571\t.\t-\t.\tgene_id "ATMG00950";\n+ChrM\tprotein_coding\tgene\t256865\t257476\t.\t-\t.\tgene_id "ATMG00960";\n+ChrM\tprotein_coding\tgene\t258045\t258398\t.\t-\t.\tgene_id "ATMG00970";\n+ChrM\tprotein_coding\tgene\t260224\t260601\t.\t-\t.\tgene_id "ATMG00980";\n+ChrM\tprotein_coding\tgene\t260647\t261006\t.\t-\t.\tgene_id "ATMG00990";\n+ChrM\tprotein_coding\tgene\t260963\t261307\t.\t-\t.\tgene_id "ATMG01000";\n+ChrM\tprotein_coding\tgene\t261105\t261461\t.\t-\t.\tgene_id "ATMG01010";\n+ChrM\tprotein_coding\tgene\t261617\t262078\t.\t-\t.\tgene_id "ATMG01020";\n+ChrM\tprotein_coding\tgene\t264113\t264433\t.\t-\t.\tgene_id "ATMG01030";\n+ChrM\tprotein_coding\tgene\t270238\t270561\t.\t-\t.\tgene_id "ATMG01040";\n+ChrM\tprotein_coding\tgene\t272944\t273423\t.\t-\t.\tgene_id "ATMG01050";\n+ChrM\tprotein_coding\tgene\t274899\t275222\t.\t+\t.\tgene_id "ATMG01060";\n+ChrM\ttRNA\tgene\t275103\t275176\t.\t+\t.\tgene_id "ATMG01070";\n+ChrM\tprotein_coding\tgene\t278649\t279152\t.\t+\t.\tgene_id "ATMG01080";\n+ChrM\tprotein_coding\tgene\t279463\t280251\t.\t+\t.\tgene_id "ATMG01090";\n+ChrM\tprotein_coding\tgene\t281614\t281931\t.\t+\t.\tgene_id "ATMG01100";\n+ChrM\tprotein_coding\tgene\t283034\t283789\t.\t-\t.\tgene_id "ATMG01110";\n+ChrM\tprotein_coding\tgene\t287917\t289083\t.\t-\t.\tgene_id "ATMG01120";\n+ChrM\tprotein_coding\tgene\t289197\t289517\t.\t-\t.\tgene_id "ATMG01130";\n+ChrM\tprotein_coding\tgene\t292973\t293431\t.\t-\t.\tgene_id "ATMG01140";\n+ChrM\tprotein_coding\tgene\t293976\t294296\t.\t-\t.\tgene_id "ATMG01150";\n+ChrM\ttRNA\tgene\t296688\t296784\t.\t-\t.\tgene_id "ATMG01160";\n+ChrM\tprotein_coding\tgene\t296820\t298204\t.\t-\t.\tgene_id "ATMG01170";\n+ChrM\tprotein_coding\tgene\t300966\t301301\t.\t-\t.\tgene_id "ATMG01180";\n+ChrM\tprotein_coding\tgene\t302166\t303689\t.\t-\t.\tgene_id "ATMG01190";\n+ChrM\tprotein_coding\tgene\t303836\t304720\t.\t-\t.\tgene_id "ATMG01200";\n+ChrM\tprotein_coding\tgene\t304779\t305084\t.\t-\t.\tgene_id "ATMG01210";\n+ChrM\tprotein_coding\tgene\t305221\t305562\t.\t-\t.\tgene_id "ATMG01220";\n+ChrM\tprotein_coding\tgene\t308932\t309369\t.\t+\t.\tgene_id "ATMG01230";\n+ChrM\tprotein_coding\tgene\t309338\t309640\t.\t-\t.\tgene_id "ATMG01240";\n+ChrM\tprotein_coding\tgene\t310514\t310882\t.\t+\t.\tgene_id "ATMG01250";\n+ChrM\tprotein_coding\tgene\t312545\t313162\t.\t-\t.\tgene_id "ATMG01260";\n+ChrM\tprotein_coding\tgene\t314627\t315073\t.\t+\t.\tgene_id "ATMG01270";\n+ChrM\tprotein_coding\tgene\t318004\t318390\t.\t-\t.\tgene_id "ATMG01275";\n+ChrM\tprotein_coding\tgene\t318588\t319463\t.\t-\t.\tgene_id "ATMG01280";\n+ChrM\tprotein_coding\tgene\t324259\t324594\t.\t+\t.\tgene_id "ATMG01290";\n+ChrM\tprotein_coding\tgene\t325300\t325710\t.\t+\t.\tgene_id "ATMG01300";\n+ChrM\tprotein_coding\tgene\t326099\t326509\t.\t+\t.\tgene_id "ATMG01310";\n+ChrM\tprotein_coding\tgene\t327890\t333105\t.\t-\t.\tgene_id "ATMG01320";\n+ChrM\tprotein_coding\tgene\t329082\t329465\t.\t-\t.\tgene_id "ATMG01330";\n+ChrM\ttRNA\tgene\t337669\t337741\t.\t+\t.\tgene_id "ATMG01340";\n+ChrM\tprotein_coding\tgene\t346757\t347194\t.\t+\t.\tgene_id "ATMG01350";\n+ChrM\tprotein_coding\tgene\t349830\t351413\t.\t-\t.\tgene_id "ATMG01360";\n+ChrM\tprotein_coding\tgene\t360717\t361052\t.\t-\t.\tgene_id "ATMG01370";\n+ChrM\trRNA\tgene\t361062\t361179\t.\t-\t.\tgene_id "ATMG01380";\n+ChrM\trRNA\tgene\t361350\t363284\t.\t-\t.\tgene_id "ATMG01390";\n+ChrM\tprotein_coding\tgene\t363725\t364042\t.\t+\t.\tgene_id "ATMG01400";\n+ChrM\tprotein_coding\tgene\t366086\t366700\t.\t-\t.\tgene_id "ATMG01410";\n'
b
diff -r 000000000000 -r 0a0bba8e1823 test-data/input_positive_pair.tsv
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/input_positive_pair.tsv Wed Feb 22 07:32:08 2017 -0500
b
b'@@ -0,0 +1,874 @@\n+AT2G46910\tAT2G46915\n+AT3G12250\tAT3G12260\n+AT5G50315\tAT5G50320\n+AT1G76780\tAT1G76790\n+AT3G10140\tAT3G10150\n+AT3G47550\tAT3G47560\n+AT3G17160\tAT3G17170\n+AT3G63490\tAT3G63500\n+AT1G24030\tAT1G24040\n+AT1G22840\tAT1G22850\n+AT2G19330\tAT2G19340\n+AT3G53240\tAT3G53235\n+AT4G29120\tAT4G29130\n+AT1G47410\tAT1G47420\n+AT3G03320\tAT3G03330\n+AT1G06450\tAT1G06460\n+AT3G21200\tAT3G21210\n+AT5G07110\tAT5G07120\n+AT1G73030\tAT1G73040\n+AT1G24600\tAT1G24610\n+AT1G64270\tAT1G64280\n+AT1G18090\tAT1G18100\n+AT2G20495\tAT2G20500\n+AT4G14700\tAT4G14710\n+AT3G07400\tAT3G07410\n+AT2G23940\tAT2G23945\n+AT5G20130\tAT5G20140\n+AT5G39970\tAT5G39980\n+AT2G29180\tAT2G29190\n+AT2G32800\tAT2G32810\n+AT4G17270\tAT4G17280\n+AT5G01715\tAT5G01720\n+AT4G16400\tAT4G16410\n+AT5G16840\tAT5G16850\n+AT5G05190\tAT5G05200\n+AT4G36720\tAT4G36730\n+AT5G18100\tAT5G18110\n+AT1G10522\tAT1G10530\n+AT1G72030\tAT1G72040\n+AT4G21160\tAT4G21170\n+AT4G35410\tAT4G35420\n+AT1G22800\tAT1G22810\n+AT1G50440\tAT1G50450\n+AT5G67530\tAT5G67540\n+AT3G07990\tAT3G08000\n+AT1G48560\tAT1G48570\n+AT3G04870\tAT3G04880\n+AT2G18340\tAT2G18350\n+AT1G07210\tAT1G07220\n+AT1G14350\tAT1G14360\n+AT5G27280\tAT5G27290\n+AT1G76690\tAT1G76700\n+AT4G28070\tAT4G28080\n+AT1G24340\tAT1G24350\n+AT2G04400\tAT2G04410\n+AT1G76050\tAT1G76060\n+AT4G38560\tAT4G38570\n+AT4G18370\tAT4G18372\n+AT3G20670\tAT3G20680\n+AT3G51390\tAT3G51400\n+AT3G26840\tAT3G26850\n+AT4G04460\tAT4G04470\n+AT5G67130\tAT5G67140\n+AT4G15720\tAT4G15730\n+AT1G79810\tAT1G79820\n+AT5G25930\tAT5G25940\n+AT1G15950\tAT1G15960\n+AT4G30480\tAT4G30490\n+AT1G28100\tAT1G28110\n+AT2G05710\tAT2G05715\n+AT5G23070\tAT5G23080\n+AT4G03180\tAT4G03190\n+AT2G03270\tAT2G03280\n+AT1G62790\tAT1G62800\n+AT5G43940\tAT5G43950\n+AT1G15740\tAT1G15750\n+AT5G41140\tAT5G41150\n+AT5G09270\tAT5G09280\n+AT5G04170\tAT5G04180\n+AT4G13580\tAT4G13590\n+AT3G28730\tAT3G28740\n+AT2G34380\tAT2G34390\n+AT3G17380\tAT3G17390\n+AT5G48730\tAT5G48740\n+AT3G56940\tAT3G56950\n+AT4G23610\tAT4G23620\n+AT2G32670\tAT2G32680\n+AT2G27460\tAT2G27470\n+AT1G75170\tAT1G75180\n+AT5G17990\tAT5G18000\n+AT5G24490\tAT5G24500\n+AT1G74430\tAT1G74440\n+AT4G26430\tAT4G26440\n+AT5G58620\tAT5G58630\n+AT4G17830\tAT4G17840\n+AT2G15900\tAT2G15910\n+AT2G45530\tAT2G45540\n+AT5G54680\tAT5G54690\n+AT5G03020\tAT5G03030\n+AT2G18280\tAT2G18270\n+AT1G54780\tAT1G54790\n+AT5G54080\tAT5G54090\n+AT4G03140\tAT4G03150\n+AT1G11530\tAT1G11540\n+AT3G55390\tAT3G55400\n+AT3G46950\tAT3G46960\n+AT1G15030\tAT1G15040\n+AT2G26920\tAT2G26930\n+AT5G64680\tAT5G64685\n+AT4G38170\tAT4G38180\n+AT3G62080\tAT3G62090\n+AT5G62670\tAT5G62680\n+AT5G10160\tAT5G10170\n+AT2G41110\tAT2G41120\n+AT3G60910\tAT3G60920\n+AT1G76260\tAT1G76270\n+AT1G22050\tAT1G22060\n+AT3G58510\tAT3G58520\n+AT1G30010\tAT1G30020\n+AT3G01920\tAT3G01930\n+AT3G47660\tAT3G47670\n+AT4G21300\tAT4G21310\n+AT4G18395\tAT4G18400\n+AT2G26640\tAT2G26650\n+AT3G49180\tAT3G49190\n+AT4G01790\tAT4G01800\n+AT2G27600\tAT2G27610\n+AT4G05450\tAT4G05460\n+AT5G61360\tAT5G61370\n+AT2G38010\tAT2G38020\n+AT1G14800\tAT1G14810\n+AT3G59110\tAT3G59120\n+AT3G24560\tAT3G24570\n+AT5G52840\tAT5G52850\n+AT2G19060\tAT2G19070\n+AT3G20730\tAT3G20740\n+AT1G54150\tAT1G54160\n+AT4G24380\tAT4G24390\n+AT1G04030\tAT1G04040\n+AT5G07830\tAT5G07840\n+AT4G20760\tAT4G20770\n+AT3G01830\tAT3G01840\n+AT2G37330\tAT2G37340\n+AT5G26710\tAT5G26700\n+AT5G01030\tAT5G01040\n+AT4G28220\tAT4G28230\n+AT5G45030\tAT5G45040\n+AT4G39235\tAT4G39240\n+AT5G47720\tAT5G47730\n+AT1G78770\tAT1G78780\n+AT5G59860\tAT5G59850\n+AT1G49960\tAT1G49970\n+AT5G67170\tAT5G67180\n+AT5G07240\tAT5G07250\n+AT5G08780\tAT5G08790\n+AT4G19650\tAT4G19645\n+AT5G65620\tAT5G65630\n+AT5G05580\tAT5G05590\n+AT3G16130\tAT3G16140\n+AT3G52210\tAT3G52220\n+AT5G50240\tAT5G50250\n+AT1G73380\tAT1G73390\n+AT2G39470\tAT2G39480\n+AT1G68935\tAT1G68940\n+AT5G08540\tAT5G08550\n+AT4G18770\tAT4G18780\n+AT5G41680\tAT5G41685\n+AT2G44750\tAT2G44760\n+AT3G11560\tAT3G11570\n+AT4G33355\tAT4G33360\n+AT2G34780\tAT2G34790\n+AT1G19710\tAT1G19715\n+AT2G43160\tAT2G43180\n+AT2G30950\tAT2G30960\n+AT4G31850\tAT4G31860\n+AT1G64185\tAT1G64190\n+AT2G42220\tAT2G42230\n+AT1G78130\tAT1G78140\n+AT3G50860\tAT3G50850\n+AT4G34080\tAT4G34090\n+AT5G58230\tAT5G58240\n+AT3G14100\tAT3G14110\n+AT5G65990\tAT5G66000\n+AT4G32560\tAT4G32570\n+AT2G22300\tAT2G22310\n+AT4G38225\tAT4G38230\n+AT3G07170\tAT3G07180\n+AT1G27060\tAT1G27070\n+AT1G31480\tAT1G31490\n+AT5G66280\tA'..b'T4G37670\n+AT5G61210\tAT5G61220\n+AT5G67300\tAT5G67310\n+AT3G17790\tAT3G17800\n+AT5G20500\tAT5G20510\n+AT2G38040\tAT2G38050\n+AT3G59030\tAT3G59040\n+AT3G16240\tAT3G16250\n+AT3G05760\tAT3G05770\n+AT2G43550\tAT2G43560\n+AT5G16150\tAT5G16160\n+AT3G60340\tAT3G60350\n+AT4G39990\tAT4G40000\n+AT4G21180\tAT4G21190\n+AT5G49945\tAT5G49950\n+AT4G16430\tAT4G16440\n+AT4G29000\tAT4G29010\n+AT4G14100\tAT4G14090\n+AT1G47330\tAT1G47340\n+AT2G01960\tAT2G01970\n+AT5G58060\tAT5G58070\n+AT1G54680\tAT1G54690\n+AT4G37040\tAT4G37050\n+AT2G22230\tAT2G22240\n+AT1G07010\tAT1G07020\n+AT2G47980\tAT2G47990\n+AT5G14230\tAT5G14240\n+AT4G35630\tAT4G35640\n+AT3G18680\tAT3G18690\n+AT4G22730\tAT4G22740\n+AT4G36370\tAT4G36380\n+AT4G15820\tAT4G15830\n+AT1G11090\tAT1G11100\n+AT5G25170\tAT5G25180\n+AT5G63010\tAT5G63020\n+AT5G01960\tAT5G01970\n+AT1G62240\tAT1G62250\n+AT1G03970\tAT1G03980\n+AT3G02840\tAT3G02850\n+AT1G52087\tAT1G52090\n+AT5G13690\tAT5G13700\n+AT5G17410\tAT5G17420\n+AT2G36010\tAT2G36020\n+AT3G01460\tAT3G01470\n+AT4G35850\tAT4G35860\n+AT5G40640\tAT5G40645\n+AT5G62810\tAT5G62820\n+AT2G29410\tAT2G29420\n+AT1G18630\tAT1G18640\n+AT5G06310\tAT5G06320\n+AT2G31880\tAT2G31890\n+AT5G08390\tAT5G08400\n+AT4G19100\tAT4G19095\n+AT2G41950\tAT2G41960\n+AT3G62580\tAT3G62590\n+AT3G24020\tAT3G24030\n+AT1G70330\tAT1G70340\n+AT4G20300\tAT4G20290\n+AT5G13390\tAT5G13400\n+AT1G79560\tAT1G79570\n+AT3G47950\tAT3G47960\n+AT5G58300\tAT5G58310\n+AT1G31220\tAT1G31230\n+AT2G26170\tAT2G26180\n+AT5G64750\tAT5G64760\n+AT5G02120\tAT5G02130\n+AT1G18880\tAT1G18890\n+AT2G01860\tAT2G01870\n+AT4G21470\tAT4G21480\n+AT1G11770\tAT1G11780\n+AT5G38600\tAT5G38610\n+AT5G15440\tAT5G15450\n+AT2G41670\tAT2G41680\n+AT1G27300\tAT1G27310\n+AT3G45970\tAT3G45980\n+AT2G28290\tAT2G28280\n+AT4G26680\tAT4G26690\n+AT4G29250\tAT4G29260\n+AT5G47435\tAT5G47440\n+AT1G21270\tAT1G21280\n+AT1G48300\tAT1G48310\n+AT3G58710\tAT3G58720\n+AT3G29320\tAT3G29330\n+AT5G16060\tAT5G16070\n+AT5G11810\tAT5G11820\n+AT1G48100\tAT1G48110\n+AT3G14270\tAT3G14280\n+AT5G16300\tAT5G16310\n+AT1G16560\tAT1G16570\n+AT1G15410\tAT1G15415\n+AT2G31035\tAT2G31040\n+AT5G26030\tAT5G26040\n+AT2G41760\tAT2G41770\n+AT5G38630\tAT5G38640\n+AT3G48320\tAT3G48330\n+AT3G27750\tAT3G27770\n+AT5G02710\tAT5G02720\n+AT1G63980\tAT1G63990\n+AT1G71180\tAT1G71190\n+AT1G17130\tAT1G17140\n+AT4G24010\tAT4G24015\n+AT1G15810\tAT1G15820\n+AT2G36580\tAT2G36590\n+AT4G34410\tAT4G34412\n+AT4G18570\tAT4G18580\n+AT1G56000\tAT1G56010\n+AT5G51010\tAT5G51020\n+AT3G61560\tAT3G61570\n+AT1G31190\tAT1G31200\n+AT1G78050\tAT1G78060\n+AT1G16810\tAT1G16820\n+AT4G36760\tAT4G36770\n+AT5G41980\tAT5G41990\n+AT5G63880\tAT5G63890\n+AT2G20260\tAT2G20270\n+AT4G04860\tAT4G04870\n+AT3G62010\tAT3G62020\n+AT5G24800\tAT5G24810\n+AT5G63480\tAT5G63490\n+AT4G33666\tAT4G33670\n+AT1G54830\tAT1G54840\n+AT5G49910\tAT5G49920\n+AT5G66180\tAT5G66190\n+AT1G27650\tAT1G27640\n+AT4G29830\tAT4G29840\n+AT3G19700\tAT3G19710\n+AT5G42180\tAT5G42190\n+AT4G28760\tAT4G28770\n+AT1G05450\tAT1G05460\n+AT2G45990\tAT2G46000\n+AT3G13670\tAT3G13672\n+AT3G54780\tAT3G54790\n+AT2G46225\tAT2G46230\n+AT4G34850\tAT4G34860\n+AT1G04210\tAT1G04220\n+AT1G77230\tAT1G77240\n+AT5G58960\tAT5G58970\n+AT4G11970\tAT4G11980\n+AT1G14000\tAT1G14010\n+AT3G15870\tAT3G15880\n+AT2G40765\tAT2G40770\n+AT1G29320\tAT1G29330\n+AT2G27080\tAT2G27090\n+AT2G39340\tAT2G39350\n+AT5G02820\tAT5G02830\n+AT1G55265\tAT1G55270\n+AT5G60510\tAT5G60520\n+AT5G41320\tAT5G41330\n+AT2G16910\tAT2G16920\n+AT1G71300\tAT1G71310\n+AT1G09000\tAT1G09010\n+AT5G40155\tAT5G40160\n+AT3G61540\tAT3G61530\n+AT2G14680\tAT2G14690\n+AT4G28040\tAT4G28050\n+AT5G06420\tAT5G06410\n+AT1G43890\tAT1G43895\n+AT1G73220\tAT1G73230\n+AT5G17630\tAT5G17640\n+AT1G65960\tAT1G65970\n+AT3G13770\tAT3G13772\n+AT4G26970\tAT4G26980\n+AT3G22620\tAT3G22630\n+AT3G55030\tAT3G55040\n+AT5G55820\tAT5G55810\n+AT4G10470\tAT4G10480\n+AT1G10110\tAT1G10120\n+AT5G47640\tAT5G47650\n+AT5G49600\tAT5G49610\n+AT3G15060\tAT3G15070\n+AT5G19840\tAT5G19850\n+AT1G76350\tAT1G76360\n+AT1G79660\tAT1G79670\n+AT1G50520\tAT1G50530\n+AT1G60420\tAT1G60430\n+AT3G25550\tAT3G25560\n+AT5G65480\tAT5G65490\n+AT3G27110\tAT3G27120\n+AT1G23730\tAT1G23740\n+AT5G13720\tAT5G13730\n+AT5G47455\tAT5G47460\n+AT4G33100\tAT4G33110\n+AT3G61220\tAT3G61230\n+AT4G10020\tAT4G10030\n+AT3G15605\tAT3G15610\n+AT1G53440\tAT1G53450\n+AT5G52560\tAT5G52570\n+AT1G74550\tAT1G74560\n+AT5G43770\tAT5G43780\n+AT5G06220\tAT5G06230\n+AT5G02280\tAT5G02290\n'
b
diff -r 000000000000 -r 0a0bba8e1823 test-data/input_read_count_asmt.tsv
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/input_read_count_asmt.tsv Wed Feb 22 07:32:08 2017 -0500
b
b'@@ -0,0 +1,33597 @@\n+AT1G38440\t0\t0\t0\n+AT1G43171\t0\t0\t0\n+AT1G67670\t0\t0\t2\n+AT1G75280\t0\t6\t8\n+AT1G03010\t0\t0\t0\n+AT1G19850\t6\t21\t15\n+AT1G41810\t0\t0\t0\n+AT1G30610\t0\t0\t0\n+AT1G14048\t0\t0\t0\n+AT1G37045\t0\t0\t0\n+AT1G37040\t0\t0\t0\n+AT1G74880\t0\t0\t0\n+AT1G06990\t1\t0\t0\n+AT1G75420\t1\t0\t5\n+AT1G70410\t7\t13\t13\n+AT1G80640\t2\t3\t3\n+AT1G41840\t0\t0\t0\n+AT1G36310\t0\t0\t1\n+AT1G74360\t3\t3\t0\n+AT1G58430\t0\t0\t0\n+AT1G73140\t0\t1\t5\n+AT1G72810\t8\t2\t0\n+AT1G01590\t0\t0\t0\n+AT1G63810\t4\t10\t0\n+AT1G68790\t0\t11\t0\n+AT1G68795\t16\t10\t11\n+AT1G48830\t5\t14\t2\n+AT1G76240\t9\t2\t10\n+AT1G67310\t0\t14\t1\n+AT1G32560\t0\t0\t0\n+AT1G35030\t0\t0\t0\n+AT1G64660\t1\t2\t1\n+AT1G35035\t0\t0\t0\n+AT1G16705\t0\t0\t0\n+AT1G69440\t0\t0\t0\n+AT1G62490\t0\t0\t1\n+AT1G78940\t3\t5\t0\n+AT1G31200\t0\t2\t1\n+AT1G71730\t0\t14\t2\n+AT1G49450\t0\t0\t0\n+AT1G45170\t0\t8\t0\n+AT1G30550\t3\t9\t9\n+AT1G65200\t0\t0\t0\n+AT1G10460\t6\t13\t2\n+AT1G46480\t0\t0\t0\n+AT1G56170\t7\t9\t2\n+AT1G62960\t6\t1\t5\n+AT1G21945\t0\t0\t0\n+AT1G21940\t0\t0\t0\n+AT1G31150\t1\t0\t0\n+AT1G11100\t65\t21\t23\n+AT1G14610\t8\t5\t10\n+AT1G64320\t0\t1\t0\n+AT1G73805\t0\t0\t0\n+AT1G19930\t0\t0\t0\n+AT1G35450\t0\t0\t0\n+AT1G67870\t0\t0\t0\n+AT1G06750\t4\t0\t0\n+AT1G18040\t4\t14\t4\n+AT1G47465\t0\t0\t1\n+AT1G26620\t9\t1\t0\n+AT1G52942\t0\t0\t0\n+AT1G52940\t0\t0\t0\n+AT1G22710\t3\t0\t3\n+AT1G32049\t0\t0\t0\n+AT1G32040\t0\t0\t0\n+AT1G32045\t0\t0\t0\n+AT1G80260\t2\t7\t1\n+AT1G01070\t4\t1\t0\n+AT1G54360\t56\t69\t56\n+AT1G12120\t3\t5\t8\n+AT1G23390\t1\t2\t0\n+AT1G29990\t4\t0\t6\n+AT1G51410\t0\t4\t5\n+AT1G29660\t0\t0\t0\n+AT1G53300\t0\t9\t2\n+AT1G57810\t0\t0\t0\n+AT1G50450\t3\t1\t1\n+AT1G07800\t0\t0\t0\n+AT1G11475\t8\t24\t6\n+AT1G79540\t0\t0\t0\n+AT1G74710\t0\t0\t0\n+AT1G47578\t2\t0\t0\n+AT1G14470\t0\t0\t0\n+AT1G79090\t1\t1\t2\n+AT1G17540\t9\t6\t3\n+AT1G21530\t4\t0\t2\n+AT1G34570\t0\t0\t0\n+AT1G02970\t1\t0\t0\n+AT1G04720\t0\t1\t0\n+AT1G60420\t7\t8\t9\n+AT1G08600\t5\t1\t5\n+AT1G15200\t0\t3\t1\n+AT1G19580\t0\t0\t3\n+AT1G14150\t0\t0\t0\n+AT1G43755\t1\t0\t0\n+AT1G43750\t0\t0\t0\n+AT1G53480\t10\t15\t4\n+AT1G56470\t0\t0\t0\n+AT1G19050\t0\t0\t0\n+AT1G64330\t8\t2\t3\n+AT1G03730\t35\t19\t10\n+AT1G63370\t0\t0\t0\n+AT1G57590\t5\t2\t4\n+AT1G49610\t0\t0\t2\n+AT1G62055\t0\t0\t0\n+AT1G44191\t0\t0\t0\n+AT1G31040\t0\t0\t0\n+AT1G30790\t0\t0\t0\n+AT1G30795\t3\t6\t4\n+AT1G68770\t0\t0\t0\n+AT1G40470\t0\t0\t0\n+AT1G06810\t0\t0\t1\n+AT1G71970\t4\t1\t7\n+AT1G12440\t3\t5\t0\n+AT1G23610\t0\t0\t0\n+AT1G02620\t0\t0\t0\n+AT1G70590\t13\t5\t13\n+AT1G69260\t27\t11\t16\n+AT1G25025\t3\t0\t0\n+AT1G78360\t0\t0\t0\n+AT1G62150\t5\t0\t0\n+AT1G63600\t0\t3\t0\n+AT1G70620\t1\t3\t9\n+AT1G47260\t0\t4\t4\n+AT1G47265\t0\t0\t0\n+AT1G72660\t1\t0\t9\n+AT1G77010\t1\t1\t1\n+AT1G36105\t0\t0\t0\n+AT1G42160\t0\t0\t0\n+AT1G28730\t0\t0\t0\n+AT1G48200\t4\t3\t0\n+AT1G68920\t5\t4\t7\n+AT1G32680\t0\t0\t0\n+AT1G22080\t2\t0\t0\n+AT1G67030\t4\t1\t0\n+AT1G33880\t0\t7\t0\n+AT1G62750\t15\t2\t7\n+AT1G67500\t9\t6\t0\n+AT1G12320\t0\t0\t0\n+AT1G15730\t26\t14\t17\n+AT1G26700\t0\t2\t5\n+AT1G60950\t1\t10\t0\n+AT1G17665\t1\t1\t0\n+AT1G17660\t0\t0\t0\n+AT1G12490\t0\t0\t0\n+AT1G49550\t0\t0\t0\n+AT1G29465\t1\t3\t4\n+AT1G29460\t0\t0\t0\n+AT1G43950\t0\t0\t0\n+AT1G62610\t9\t2\t10\n+AT1G63320\t0\t0\t0\n+AT1G41743\t0\t0\t0\n+AT1G27070\t10\t2\t9\n+AT1G41746\t0\t0\t0\n+AT1G65900\t4\t1\t5\n+AT1G65907\t0\t0\t0\n+AT1G30960\t0\t0\t1\n+AT1G71500\t1\t1\t0\n+AT1G13130\t0\t0\t0\n+AT1G80840\t30\t8\t15\n+AT1G20930\t4\t3\t3\n+AT1G06310\t9\t0\t0\n+AT1G08440\t0\t0\t0\n+AT1G21040\t0\t0\t0\n+AT1G60620\t0\t0\t0\n+AT1G60625\t0\t0\t0\n+AT1G60990\t19\t2\t8\n+AT1G60995\t0\t0\t4\n+AT1G19210\t0\t0\t0\n+AT1G66590\t0\t0\t0\n+AT1G53970\t0\t0\t0\n+AT1G25340\t0\t0\t0\n+AT1G12860\t0\t0\t0\n+AT1G58808\t0\t0\t0\n+AT1G58807\t0\t0\t0\n+AT1G17300\t0\t0\t0\n+AT1G41893\t0\t0\t0\n+AT1G28470\t3\t0\t0\n+AT1G80540\t0\t0\t0\n+AT1G30890\t11\t1\t3\n+AT1G01360\t7\t1\t1\n+AT1G54460\t1\t4\t7\n+AT1G56270\t0\t0\t0\n+AT1G65250\t0\t0\t0\n+AT1G23580\t0\t0\t0\n+AT1G75670\t0\t0\t4\n+AT1G11880\t4\t4\t4\n+AT1G47420\t14\t13\t7\n+AT1G07950\t0\t1\t0\n+AT1G53110\t2\t0\t0\n+AT1G35970\t0\t0\t7\n+AT1G57600\t0\t0\t1\n+AT1G20580\t1\t5\t5\n+AT1G74100\t1\t12\t8\n+AT1G73730\t0\t1\t0\n+AT1G69100\t0\t0\t0\n+AT1G04390\t1\t1\t0\n+AT1G52640\t0\t0\t0\n+AT1G42367\t0\t0\t0\n+AT1G42365\t0\t0\t0\n+AT1G42360\t0\t0\t0\n+AT1G62450\t0\t0\t0\n+AT1G43090\t0\t0\t0\n+AT1G15430\t5\t1\t0\n+AT1G78750\t0\t0\t0\n+AT1G32910\t0\t0\t0\n+AT1G23770\t0\t0\t0\n+AT1G16700\t0\t5\t3\n+AT1G05440\t0\t0\t0\n+AT1G49790\t0\t0\t0\n+AT1G09140\t10\t7\t5\n+AT1G76610\t0\t0\t0\n+AT1G11735\t0\t0\t0\n+AT1G44010\t0\t0\t0\n+AT1G59810\t0\t0\t0\n+AT1G80890\t4\t0\t0\n+AT1G63360\t0\t0\t0\n+AT1G74920\t16\t8\t9\n+AT1G74929\t0\t0\t0\n+AT1G06137\t0\t0\t0\n+AT1G06135\t0\t0\t2\n+AT1G31580\t0\t0\t0\n+AT1G48500\t0\t0\t0\n+AT1G59630\t0\t3\t0\n+AT1G30180\t0\t0\t0\n+AT1G34400\t0\t0\t0\n+AT1G34405\t0\t0\t0\n+AT1G24230\t0\t0\t0\n+AT1G68070\t0\t0\t9\n+AT1G48040\t4\t0\t0\n+AT'..b'+ATCG01200\t0\t0\t0\n+ATCG00310\t0\t0\t0\n+ATCG01040\t8\t11\t22\n+ATCG00880\t0\t0\t0\n+ATCG00050\t1\t14\t6\n+ATCG00290\t0\t0\t0\n+ATCG00480\t14\t5\t3\n+ATCG00730\t0\t0\t7\n+ATCG01280\t0\t0\t0\n+ATCG00800\t3\t7\t4\n+ATCG00210\t0\t0\t0\n+ATCG01160\t0\t0\t0\n+ATCG00670\t10\t19\t32\n+ATCG00930\t0\t0\t0\n+ATCG00160\t3\t3\t1\n+ATCG01070\t4\t3\t5\n+ATCG00530\t2\t2\t2\n+ATCG00700\t3\t0\t5\n+ATCG01250\t0\t0\t0\n+ATCG01010\t36\t20\t24\n+ATCG00870\t0\t0\t0\n+ATCG00240\t0\t0\t0\n+ATCG01190\t0\t0\t0\n+ATCG00780\t0\t3\t5\n+ATCG00080\t0\t0\t0\n+ATCG00620\t0\t0\t0\n+ATCG01110\t0\t0\t0\n+ATCG00470\t21\t13\t18\n+ATCG00130\t10\t0\t3\n+ATCG00540\t1\t3\t4\n+ATCG00940\t0\t0\t0\n+ATCG01260\t0\t0\t0\n+ATCG00330\t0\t0\t0\n+ATCG00070\t2\t7\t3\n+ATCG00270\t1\t6\t0\n+ATCG00750\t4\t3\t0\n+ATCG00820\t3\t5\t1\n+ATCG01300\t0\t0\t0\n+ATCG01140\t0\t0\t0\n+ATCG00420\t0\t0\t0\n+ATCG00990\t0\t0\t0\n+ATCG01060\t2\t0\t0\n+ATCG00650\t13\t0\t0\n+ATCG00910\t0\t0\t0\n+ATCG00140\t0\t0\t1\n+ATCG01230\t0\t0\t0\n+ATCG00300\t0\t0\t3\n+ATCG00890\t0\t0\t0\n+ATCG00510\t0\t0\t0\n+ATCG00490\t4\t9\t1\n+ATCG00720\t7\t14\t18\n+ATCG01030\t0\t0\t0\n+ATCG00810\t4\t3\t2\n+ATCG00020\t2\t0\t12\n+ATCG00220\t0\t1\t4\n+ATCG00590\t2\t5\t2\n+ATCG01170\t0\t0\t0\n+ATCG00410\t0\t0\t0\n+ATCG00360\t0\t0\t12\n+ATCG01090\t3\t4\t4\n+ATCG00600\t0\t1\t1\n+ATCG00920\t0\t0\t0\n+ATCG00110\t3\t17\t12\n+ATCG00350\t5\t9\t3\n+ATCG00680\t1\t8\t0\n+ATCG00520\t0\t0\t1\n+ATCG00190\t0\t5\t3\n+ATCG01240\t0\t0\t0\n+ATCG01000\t0\t0\t0\n+ATCG00840\t0\t0\t0\n+ATCG01180\t0\t0\t0\n+ATCG00010\t3\t5\t2\n+ATCG00250\t0\t0\t0\n+ATCG00770\t3\t3\t0\n+ATCG00090\t0\t0\t0\n+ATCG00630\t5\t6\t16\n+ATCG01120\t1\t0\t1\n+ATCG00440\t3\t4\t0\n+ATCG00120\t2\t3\t2\n+ATCG00570\t0\t0\t0\n+ATCG00970\t0\t0\t0\n+ATCG01210\t0\t0\t0\n+ATCG00320\t0\t0\t0\n+ATCG01050\t6\t2\t3\n+ATCG00280\t0\t3\t0\n+ATCG00040\t12\t7\t12\n+ATMG01170\t3\t7\t1\n+ATMG01090\t0\t0\t0\n+ATMG00100\t0\t0\t0\n+ATMG00920\t0\t0\t0\n+ATMG01270\t0\t0\t0\n+ATMG00710\t0\t0\t0\n+ATMG01040\t0\t0\t0\n+ATMG00690\t3\t0\t0\n+ATMG00520\t0\t0\t0\n+ATMG00300\t0\t0\t0\n+ATMG00020\t322\t309\t284\n+ATMG00790\t0\t0\t0\n+ATMG00800\t0\t0\t0\n+ATMG00610\t0\t0\t0\n+ATMG01250\t0\t0\t1\n+ATMG00380\t0\t0\t0\n+ATMG00400\t1\t0\t0\n+ATMG00880\t0\t0\t0\n+ATMG00260\t0\t0\t0\n+ATMG01330\t0\t0\t0\n+ATMG01120\t0\t0\t0\n+ATMG00150\t10\t6\t15\n+ATMG00480\t0\t0\t0\n+ATMG00970\t0\t0\t0\n+ATMG00720\t0\t0\t1\n+ATMG00050\t0\t4\t2\n+ATMG01010\t0\t0\t0\n+ATMG00830\t0\t0\t0\n+ATMG00570\t0\t0\t0\n+ATMG01190\t0\t0\t0\n+ATMG00350\t0\t0\t0\n+ATMG01400\t5\t6\t10\n+ATMG00430\t0\t0\t0\n+ATMG00250\t0\t0\t0\n+ATMG00640\t12\t0\t2\n+ATMG01110\t0\t0\t0\n+ATMG01140\t0\t0\t0\n+ATMG00940\t0\t0\t0\n+ATMG01260\t4\t10\t0\n+ATMG00770\t0\t0\t0\n+ATMG00120\t3\t0\t0\n+ATMG01380\t7\t4\t4\n+ATMG00080\t2\t0\t0\n+ATMG01020\t0\t0\t0\n+ATMG00860\t0\t0\t0\n+ATMG01130\t3\t5\t0\n+ATMG00540\t0\t0\t0\n+ATMG00360\t0\t0\t0\n+ATMG00630\t2\t0\t3\n+ATMG01275\t0\t0\t0\n+ATMG00460\t0\t0\t0\n+ATMG00200\t0\t0\t0\n+ATMG00170\t0\t0\t0\n+ATMG00910\t0\t0\t0\n+ATMG00700\t0\t0\t0\n+ATMG00285\t0\t2\t0\n+ATMG00280\t0\t0\t0\n+ATMG01070\t0\t0\t0\n+ATMG00513\t0\t0\t0\n+ATMG00510\t0\t0\t0\n+ATMG00516\t0\t0\t1\n+ATMG00990\t0\t0\t0\n+ATMG00330\t0\t0\t0\n+ATMG00340\t0\t0\t0\n+ATMG00030\t3181\t6466\t15224\n+ATMG00780\t0\t0\t0\n+ATMG00810\t0\t0\t2\n+ATMG01300\t0\t0\t0\n+ATMG00590\t0\t0\t0\n+ATMG01150\t0\t0\t0\n+ATMG00890\t0\t0\t0\n+ATMG00270\t0\t0\t0\n+ATMG00665\t0\t0\t0\n+ATMG00660\t1\t0\t2\n+ATMG00140\t8\t48\t28\n+ATMG00490\t0\t0\t0\n+ATMG00960\t0\t0\t0\n+ATMG00750\t0\t0\t0\n+ATMG00060\t0\t5\t1\n+ATMG01000\t0\t0\t0\n+ATMG00840\t0\t0\t0\n+ATMG00560\t2\t0\t0\n+ATMG01210\t0\t0\t0\n+ATMG01080\t0\t1\t0\n+ATMG01200\t0\t0\t0\n+ATMG00650\t2\t1\t0\n+ATMG01290\t0\t0\t0\n+ATMG00190\t0\t0\t0\n+ATMG00440\t0\t0\t0\n+ATMG00220\t0\t0\t0\n+ATMG00760\t0\t0\t0\n+ATMG01160\t0\t0\t0\n+ATMG00110\t12\t3\t9\n+ATMG01390\t57\t37\t53\n+ATMG00930\t0\t0\t0\n+ATMG00090\t5\t1\t17\n+ATMG00870\t0\t0\t0\n+ATMG01310\t0\t0\t0\n+ATMG00680\t0\t0\t0\n+ATMG00530\t0\t0\t0\n+ATMG00310\t0\t0\t0\n+ATMG00010\t0\t0\t1\n+ATMG01240\t0\t0\t0\n+ATMG01220\t0\t0\t0\n+ATMG00600\t0\t0\t0\n+ATMG00390\t0\t0\t0\n+ATMG00470\t0\t0\t0\n+ATMG00210\t0\t0\t0\n+ATMG00160\t3\t0\t0\n+ATMG01340\t0\t0\t0\n+ATMG00900\t0\t0\t0\n+ATMG01350\t0\t0\t0\n+ATMG00290\t0\t0\t0\n+ATMG00730\t2\t0\t3\n+ATMG01060\t0\t0\t0\n+ATMG00410\t1\t0\t0\n+ATMG00500\t0\t0\t0\n+ATMG00980\t0\t0\t0\n+ATMG00320\t0\t0\t0\n+ATMG00040\t0\t0\t0\n+ATMG00820\t0\t0\t0\n+ATMG00580\t0\t2\t0\n+ATMG01180\t0\t0\t0\n+ATMG01230\t0\t0\t0\n+ATMG01410\t0\t0\t0\n+ATMG00420\t0\t0\t0\n+ATMG00240\t0\t0\t0\n+ATMG00670\t0\t0\t0\n+ATMG01100\t0\t0\t0\n+ATMG01370\t0\t0\t0\n+ATMG00950\t0\t0\t0\n+ATMG00740\t0\t0\t0\n+ATMG01320\t0\t0\t0\n+ATMG00130\t25\t46\t18\n+ATMG00070\t0\t1\t2\n+ATMG01030\t0\t0\t0\n+ATMG00850\t0\t0\t0\n+ATMG00550\t0\t0\t0\n+ATMG01360\t3\t6\t9\n+ATMG00370\t0\t0\t0\n+ATMG01050\t0\t0\t0\n+ATMG00620\t0\t0\t0\n+ATMG01280\t0\t0\t0\n+ATMG00180\t0\t0\t0\n+ATMG00450\t0\t0\t0\n+ATMG00230\t0\t0\t0\n'
b
diff -r 000000000000 -r 0a0bba8e1823 test-data/input_read_count_smt.tsv
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/input_read_count_smt.tsv Wed Feb 22 07:32:08 2017 -0500
b
b'@@ -0,0 +1,33597 @@\n+AT1G38440\t0\t2\t0\n+AT1G43171\t2\t8\t1\n+AT1G67670\t3\t7\t0\n+AT1G75280\t341\t171\t142\n+AT1G03010\t56\t19\t25\n+AT1G19850\t747\t357\t473\n+AT1G41810\t0\t0\t0\n+AT1G30610\t88\t93\t56\n+AT1G14048\t1\t0\t0\n+AT1G37045\t0\t0\t0\n+AT1G37040\t3\t0\t0\n+AT1G74880\t1\t13\t1\n+AT1G06990\t6\t3\t7\n+AT1G75420\t211\t216\t193\n+AT1G70410\t2647\t1401\t1956\n+AT1G80640\t140\t87\t73\n+AT1G41840\t0\t0\t0\n+AT1G36310\t823\t926\t899\n+AT1G74360\t28\t16\t44\n+AT1G58430\t0\t6\t4\n+AT1G73140\t13\t5\t20\n+AT1G72810\t189\t165\t101\n+AT1G01590\t0\t0\t0\n+AT1G63810\t340\t398\t337\n+AT1G68790\t564\t870\t526\n+AT1G68795\t59\t4\t22\n+AT1G48830\t1746\t1611\t1574\n+AT1G76240\t587\t329\t460\n+AT1G67310\t342\t251\t379\n+AT1G32560\t55\t10\t22\n+AT1G35030\t1\t0\t0\n+AT1G64660\t225\t150\t110\n+AT1G35035\t0\t0\t0\n+AT1G16705\t8\t4\t11\n+AT1G69440\t10\t10\t10\n+AT1G62490\t0\t0\t0\n+AT1G78940\t27\t14\t20\n+AT1G31200\t3\t3\t0\n+AT1G71730\t227\t187\t189\n+AT1G49450\t114\t62\t79\n+AT1G45170\t257\t253\t201\n+AT1G30550\t335\t233\t225\n+AT1G65200\t1\t3\t0\n+AT1G10460\t16\t1\t3\n+AT1G46480\t2\t0\t0\n+AT1G56170\t90\t57\t58\n+AT1G62960\t91\t117\t101\n+AT1G21945\t0\t0\t0\n+AT1G21940\t0\t1\t0\n+AT1G31150\t80\t99\t123\n+AT1G11100\t161\t27\t111\n+AT1G14610\t1044\t1250\t1253\n+AT1G64320\t24\t26\t40\n+AT1G73805\t0\t5\t11\n+AT1G19930\t0\t0\t0\n+AT1G35450\t1\t0\t0\n+AT1G67870\t9\t1\t2\n+AT1G06750\t60\t54\t30\n+AT1G18040\t94\t111\t83\n+AT1G47465\t0\t0\t0\n+AT1G26620\t126\t87\t69\n+AT1G52942\t0\t0\t0\n+AT1G52940\t0\t0\t0\n+AT1G22710\t342\t118\t154\n+AT1G32049\t3\t0\t0\n+AT1G32040\t0\t0\t0\n+AT1G32045\t0\t0\t0\n+AT1G80260\t290\t317\t288\n+AT1G01070\t90\t62\t53\n+AT1G54360\t95\t82\t98\n+AT1G12120\t409\t286\t389\n+AT1G23390\t103\t63\t91\n+AT1G29990\t553\t543\t597\n+AT1G51410\t0\t0\t0\n+AT1G29660\t0\t0\t0\n+AT1G53300\t748\t621\t764\n+AT1G57810\t4\t5\t0\n+AT1G50450\t81\t58\t37\n+AT1G07800\t0\t1\t0\n+AT1G11475\t198\t184\t178\n+AT1G79540\t194\t185\t208\n+AT1G74710\t101\t107\t129\n+AT1G47578\t21\t20\t28\n+AT1G14470\t37\t38\t17\n+AT1G79090\t467\t495\t456\n+AT1G17540\t0\t0\t3\n+AT1G21530\t8\t38\t9\n+AT1G34570\t27\t51\t31\n+AT1G02970\t80\t69\t66\n+AT1G04720\t0\t0\t0\n+AT1G60420\t413\t380\t352\n+AT1G08600\t463\t570\t483\n+AT1G15200\t625\t752\t535\n+AT1G19580\t649\t606\t618\n+AT1G14150\t0\t2\t0\n+AT1G43755\t0\t0\t0\n+AT1G43750\t0\t0\t0\n+AT1G53480\t876\t774\t716\n+AT1G56470\t0\t3\t8\n+AT1G19050\t35\t18\t2\n+AT1G64330\t1634\t2270\t2193\n+AT1G03730\t473\t217\t217\n+AT1G63370\t1\t0\t7\n+AT1G57590\t659\t573\t780\n+AT1G49610\t0\t0\t1\n+AT1G62055\t0\t0\t0\n+AT1G44191\t0\t0\t0\n+AT1G31040\t3\t0\t2\n+AT1G30790\t0\t0\t0\n+AT1G30795\t0\t0\t0\n+AT1G68770\t10\t11\t6\n+AT1G40470\t0\t0\t1\n+AT1G06810\t0\t0\t0\n+AT1G71970\t49\t35\t35\n+AT1G12440\t563\t376\t384\n+AT1G23610\t0\t0\t0\n+AT1G02620\t0\t0\t0\n+AT1G70590\t239\t134\t309\n+AT1G69260\t730\t350\t428\n+AT1G25025\t0\t0\t0\n+AT1G78360\t0\t0\t0\n+AT1G62150\t221\t185\t157\n+AT1G63600\t1\t0\t0\n+AT1G70620\t241\t190\t188\n+AT1G47260\t1982\t1783\t2035\n+AT1G47265\t0\t0\t0\n+AT1G72660\t24\t5\t14\n+AT1G77010\t72\t50\t46\n+AT1G36105\t0\t0\t0\n+AT1G42160\t0\t0\t0\n+AT1G28730\t0\t0\t0\n+AT1G48200\t203\t195\t253\n+AT1G68920\t546\t328\t393\n+AT1G32680\t0\t0\t0\n+AT1G22080\t0\t0\t0\n+AT1G67030\t59\t55\t67\n+AT1G33880\t0\t0\t0\n+AT1G62750\t172\t191\t178\n+AT1G67500\t152\t116\t104\n+AT1G12320\t18\t14\t13\n+AT1G15730\t54\t29\t31\n+AT1G26700\t154\t150\t294\n+AT1G60950\t307\t209\t191\n+AT1G17665\t12\t13\t10\n+AT1G17660\t0\t0\t0\n+AT1G12490\t0\t0\t0\n+AT1G49550\t9\t8\t0\n+AT1G29465\t37\t19\t14\n+AT1G29460\t0\t0\t4\n+AT1G43950\t1\t0\t0\n+AT1G62610\t51\t97\t43\n+AT1G63320\t0\t0\t0\n+AT1G41743\t0\t0\t0\n+AT1G27070\t199\t163\t187\n+AT1G41746\t0\t0\t0\n+AT1G65900\t57\t60\t41\n+AT1G65907\t0\t0\t0\n+AT1G30960\t157\t140\t123\n+AT1G71500\t133\t34\t92\n+AT1G13130\t0\t4\t0\n+AT1G80840\t2110\t763\t1390\n+AT1G20930\t74\t80\t56\n+AT1G06310\t53\t59\t49\n+AT1G08440\t0\t0\t3\n+AT1G21040\t0\t0\t0\n+AT1G60620\t190\t149\t152\n+AT1G60625\t0\t0\t0\n+AT1G60990\t59\t41\t31\n+AT1G60995\t207\t143\t221\n+AT1G19210\t204\t192\t215\n+AT1G66590\t29\t32\t15\n+AT1G53970\t0\t0\t0\n+AT1G25340\t0\t0\t0\n+AT1G12860\t9\t12\t1\n+AT1G58808\t0\t0\t0\n+AT1G58807\t0\t0\t0\n+AT1G17300\t59\t49\t92\n+AT1G41893\t0\t0\t0\n+AT1G28470\t77\t25\t36\n+AT1G80540\t0\t0\t0\n+AT1G30890\t593\t472\t557\n+AT1G01360\t360\t155\t293\n+AT1G54460\t514\t516\t501\n+AT1G56270\t0\t0\t0\n+AT1G65250\t14\t6\t10\n+AT1G23580\t0\t0\t0\n+AT1G75670\t196\t260\t206\n+AT1G11880\t70\t63\t66\n+AT1G47420\t3536\t2402\t3386\n+AT1G07950\t122\t134\t128\n+AT1G53110\t112\t91\t88\n+AT1G35970\t0\t0\t0\n+AT1G57600\t331\t375\t390\n+AT1G20580\t931\t763\t638\n+AT1G74100\t737\t492\t560\n+AT1G73730\t157\t119\t158\n+AT1G69100\t0\t0\t0\n+AT1G04390\t96\t117\t116\n+AT1G52640\t37\t73\t36\n+AT1G42367\t0\t0\t0\n+AT1G42365\t0\t0\t0\n+AT1G42360\t0'..b'800\t424\t813\t827\n+ATCG00210\t0\t0\t0\n+ATCG01160\t0\t0\t0\n+ATCG00670\t693\t1044\t1255\n+ATCG00930\t0\t0\t0\n+ATCG00160\t67\t172\t143\n+ATCG01070\t7\t9\t9\n+ATCG00530\t26\t29\t34\n+ATCG00700\t6\t0\t8\n+ATCG01250\t0\t0\t0\n+ATCG01010\t3\t10\t24\n+ATCG00870\t0\t0\t0\n+ATCG00240\t0\t0\t0\n+ATCG01190\t0\t0\t0\n+ATCG00780\t140\t333\t352\n+ATCG00080\t18\t5\t3\n+ATCG00620\t0\t0\t12\n+ATCG01110\t6\t17\t41\n+ATCG00470\t1107\t1534\t1358\n+ATCG00130\t59\t224\t277\n+ATCG00540\t288\t506\t784\n+ATCG00940\t0\t0\t0\n+ATCG01260\t0\t0\t0\n+ATCG00330\t6\t29\t21\n+ATCG00070\t4\t0\t0\n+ATCG00270\t2\t0\t2\n+ATCG00750\t165\t230\t222\n+ATCG00820\t220\t198\t232\n+ATCG01300\t0\t0\t0\n+ATCG01140\t0\t0\t0\n+ATCG00420\t5\t7\t14\n+ATCG00990\t0\t0\t0\n+ATCG01060\t0\t1\t2\n+ATCG00650\t9\t8\t4\n+ATCG00910\t0\t0\t0\n+ATCG00140\t10\t27\t44\n+ATCG01230\t0\t0\t0\n+ATCG00300\t56\t44\t96\n+ATCG00890\t0\t0\t0\n+ATCG00510\t1\t1\t0\n+ATCG00490\t22\t55\t50\n+ATCG00720\t277\t486\t553\n+ATCG01030\t0\t0\t0\n+ATCG00810\t67\t136\t139\n+ATCG00020\t18\t62\t82\n+ATCG00220\t61\t37\t46\n+ATCG00590\t0\t1\t0\n+ATCG01170\t0\t0\t0\n+ATCG00410\t0\t2\t6\n+ATCG00360\t23\t56\t85\n+ATCG01090\t38\t54\t125\n+ATCG00600\t0\t0\t0\n+ATCG00920\t0\t0\t0\n+ATCG00110\t0\t0\t0\n+ATCG00350\t28\t32\t36\n+ATCG00680\t7\t15\t14\n+ATCG00520\t37\t31\t43\n+ATCG00190\t10\t33\t11\n+ATCG01240\t0\t0\t1\n+ATCG01000\t0\t0\t0\n+ATCG00840\t0\t0\t0\n+ATCG01180\t0\t0\t0\n+ATCG00010\t0\t0\t0\n+ATCG00250\t0\t0\t2\n+ATCG00770\t126\t213\t216\n+ATCG00090\t17\t30\t19\n+ATCG00630\t0\t0\t3\n+ATCG01120\t7\t7\t8\n+ATCG00440\t102\t96\t174\n+ATCG00120\t28\t83\t91\n+ATCG00570\t0\t0\t0\n+ATCG00970\t0\t0\t0\n+ATCG01210\t0\t1\t0\n+ATCG00320\t2\t0\t0\n+ATCG01050\t14\t4\t2\n+ATCG00280\t20\t5\t3\n+ATCG00040\t57\t277\t215\n+ATMG01170\t6\t1\t11\n+ATMG01090\t0\t0\t0\n+ATMG00100\t0\t0\t0\n+ATMG00920\t0\t0\t0\n+ATMG01270\t0\t0\t0\n+ATMG00710\t0\t0\t0\n+ATMG01040\t0\t0\t0\n+ATMG00690\t30\t50\t40\n+ATMG00520\t0\t0\t1\n+ATMG00300\t0\t0\t0\n+ATMG00020\t31242\t31736\t38893\n+ATMG00790\t0\t0\t0\n+ATMG00800\t0\t0\t0\n+ATMG00610\t0\t0\t0\n+ATMG01250\t0\t0\t0\n+ATMG00380\t0\t0\t0\n+ATMG00400\t0\t0\t0\n+ATMG00880\t0\t0\t0\n+ATMG00260\t0\t0\t0\n+ATMG01330\t0\t0\t1\n+ATMG01120\t16\t33\t19\n+ATMG00150\t0\t0\t0\n+ATMG00480\t0\t0\t0\n+ATMG00970\t0\t1\t0\n+ATMG00720\t0\t0\t0\n+ATMG00050\t0\t0\t0\n+ATMG01010\t0\t6\t13\n+ATMG00830\t1\t4\t0\n+ATMG00570\t1\t0\t0\n+ATMG01190\t0\t0\t3\n+ATMG00350\t0\t0\t0\n+ATMG01400\t0\t0\t1\n+ATMG00430\t0\t0\t0\n+ATMG00250\t0\t0\t0\n+ATMG00640\t9\t17\t41\n+ATMG01110\t0\t0\t1\n+ATMG01140\t0\t0\t0\n+ATMG00940\t0\t0\t0\n+ATMG01260\t0\t0\t0\n+ATMG00770\t0\t0\t0\n+ATMG00120\t1\t0\t1\n+ATMG01380\t14\t9\t19\n+ATMG00080\t52\t114\t114\n+ATMG01020\t0\t0\t0\n+ATMG00860\t0\t0\t0\n+ATMG01130\t0\t4\t0\n+ATMG00540\t0\t0\t0\n+ATMG00360\t0\t0\t0\n+ATMG00630\t0\t15\t18\n+ATMG01275\t0\t0\t0\n+ATMG00460\t0\t0\t0\n+ATMG00200\t0\t0\t0\n+ATMG00170\t0\t0\t0\n+ATMG00910\t0\t0\t0\n+ATMG00700\t0\t0\t0\n+ATMG00285\t1\t0\t1\n+ATMG00280\t0\t0\t0\n+ATMG01070\t0\t0\t0\n+ATMG00513\t0\t2\t0\n+ATMG00510\t2\t2\t1\n+ATMG00516\t6\t0\t8\n+ATMG00990\t2\t1\t7\n+ATMG00330\t0\t0\t0\n+ATMG00340\t0\t0\t0\n+ATMG00030\t13\t7\t0\n+ATMG00780\t0\t0\t0\n+ATMG00810\t0\t0\t0\n+ATMG01300\t0\t0\t0\n+ATMG00590\t0\t0\t0\n+ATMG01150\t0\t0\t0\n+ATMG00890\t0\t0\t0\n+ATMG00270\t0\t0\t0\n+ATMG00665\t2\t2\t1\n+ATMG00660\t19\t54\t58\n+ATMG00140\t0\t0\t2\n+ATMG00490\t0\t0\t0\n+ATMG00960\t0\t0\t1\n+ATMG00750\t0\t0\t0\n+ATMG00060\t10\t13\t12\n+ATMG01000\t7\t7\t30\n+ATMG00840\t0\t0\t0\n+ATMG00560\t10\t19\t29\n+ATMG01210\t0\t0\t0\n+ATMG01080\t16\t25\t47\n+ATMG01200\t0\t1\t3\n+ATMG00650\t4\t4\t1\n+ATMG01290\t0\t0\t0\n+ATMG00190\t0\t0\t0\n+ATMG00440\t0\t0\t0\n+ATMG00220\t0\t0\t0\n+ATMG00760\t0\t0\t0\n+ATMG01160\t0\t0\t0\n+ATMG00110\t4\t5\t8\n+ATMG01390\t1722\t1391\t1342\n+ATMG00930\t0\t0\t0\n+ATMG00090\t185\t759\t777\n+ATMG00870\t0\t0\t0\n+ATMG01310\t0\t0\t0\n+ATMG00680\t0\t0\t0\n+ATMG00530\t0\t0\t0\n+ATMG00310\t0\t0\t0\n+ATMG00010\t0\t0\t0\n+ATMG01240\t0\t0\t0\n+ATMG01220\t2\t0\t0\n+ATMG00600\t0\t0\t0\n+ATMG00390\t0\t0\t0\n+ATMG00470\t0\t0\t0\n+ATMG00210\t0\t0\t0\n+ATMG00160\t30\t27\t33\n+ATMG01340\t0\t0\t0\n+ATMG00900\t0\t0\t0\n+ATMG01350\t0\t0\t0\n+ATMG00290\t0\t0\t0\n+ATMG00730\t8\t0\t0\n+ATMG01060\t0\t0\t0\n+ATMG00410\t0\t0\t0\n+ATMG00500\t0\t0\t0\n+ATMG00980\t0\t0\t1\n+ATMG00320\t0\t0\t0\n+ATMG00040\t2\t4\t2\n+ATMG00820\t0\t3\t0\n+ATMG00580\t10\t5\t13\n+ATMG01180\t4\t1\t0\n+ATMG01230\t0\t0\t0\n+ATMG01410\t0\t0\t5\n+ATMG00420\t0\t0\t0\n+ATMG00240\t0\t0\t1\n+ATMG00670\t2\t3\t14\n+ATMG01100\t0\t0\t0\n+ATMG01370\t4\t11\t26\n+ATMG00950\t0\t0\t0\n+ATMG00740\t0\t0\t0\n+ATMG01320\t13\t16\t20\n+ATMG00130\t0\t0\t0\n+ATMG00070\t33\t57\t74\n+ATMG01030\t0\t0\t0\n+ATMG00850\t0\t0\t0\n+ATMG00550\t0\t0\t0\n+ATMG01360\t32\t81\t129\n+ATMG00370\t0\t0\t0\n+ATMG01050\t0\t0\t0\n+ATMG00620\t0\t0\t0\n+ATMG01280\t0\t0\t0\n+ATMG00180\t0\t0\t0\n+ATMG00450\t0\t0\t0\n+ATMG00230\t0\t0\t0\n'
b
diff -r 000000000000 -r 0a0bba8e1823 test-data/output_neworphan.tsv
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/output_neworphan.tsv Wed Feb 22 07:32:08 2017 -0500
b
@@ -0,0 +1,414 @@
+ATMG00030
+AT5G49440
+AT2G11240
+AT5G49640
+AT2G33320
+AT2G24840
+AT4G17905
+AT1G72760
+AT1G21310
+AT3G23460
+AT3G53510
+AT3G04690
+AT5G13560
+AT3G22800
+AT4G15150
+AT5G59580
+AT1G55180
+AT1G19220
+AT3G41768
+AT1G68945
+AT5G01560
+AT4G08850
+AT5G19100
+AT1G29540
+AT3G19020
+AT2G34330
+ATMG00130
+AT5G51580
+AT5G62140
+AT1G25450
+ATMG00140
+AT5G16770
+AT2G15320
+AT3G44798
+AT1G23840
+AT4G14860
+AT4G19170
+AT1G79910
+AT3G06433
+AT2G38823
+AT2G07688
+AT2G38210
+AT1G22880
+AT2G21188
+AT5G60280
+AT3G56780
+AT3G32980
+AT3G54590
+AT3G28550
+AT4G21740
+AT3G57400
+AT2G46810
+AT1G04540
+AT2G39380
+AT1G35710
+AT5G66330
+AT5G09530
+AT2G42860
+AT5G37760
+AT3G25010
+AT3G53720
+AT2G35000
+AT1G03940
+AT5G25010
+AT1G49780
+AT3G04440
+AT5G01930
+AT5G24318
+AT1G73330
+AT2G38830
+AT1G63860
+AT3G14650
+AT1G65346
+AT1G66690
+AT5G67460
+AT5G26960
+AT4G08400
+AT5G52020
+AT4G37620
+AT2G05440
+AT5G53740
+AT4G02320
+AT1G03495
+ATMG00150
+AT5G22890
+AT5G01170
+AT1G41930
+AT1G14070
+AT5G55490
+AT3G14660
+AT4G03270
+AT4G26670
+AT2G37100
+AT3G60080
+AT5G57180
+AT3G28630
+AT2G30505
+ATCG00630
+AT3G25011
+AT5G58530
+AT2G02700
+AT4G31710
+AT1G29770
+AT3G09070
+AT3G49520
+AT3G14950
+AT5G40900
+AT1G63140
+AT5G27100
+AT3G46910
+AT3G14630
+AT5G49700
+AT1G35580
+AT5G58170
+AT2G35960
+AT3G17640
+AT5G15900
+AT4G29930
+AT5G22610
+AT2G47720
+AT1G78030
+AT1G73880
+AT1G06148
+AT3G53950
+AT1G10260
+AT5G58080
+ATCG01100
+AT3G26932
+AT4G02520
+AT1G66480
+AT1G68795
+AT4G15975
+AT4G08150
+AT3G27025
+AT5G09360
+AT2G37380
+AT4G34220
+AT3G11090
+AT1G07340
+AT2G17064
+ATMG00110
+AT4G01720
+AT4G38300
+AT1G28400
+AT5G44569
+AT3G13760
+AT5G42750
+AT1G49490
+AT4G23670
+ATCG01080
+AT1G23720
+AT5G42040
+AT2G31990
+ATCG00050
+ATCG00380
+AT1G73260
+AT5G57750
+AT5G36870
+AT1G77640
+AT1G75820
+AT4G33560
+AT2G01320
+AT1G73066
+AT1G21990
+AT3G56810
+ATCG00065
+AT3G05820
+AT2G23540
+AT5G25020
+AT3G01420
+AT1G26890
+AT1G13150
+AT4G33470
+AT4G38460
+AT3G46550
+AT3G25020
+AT3G50350
+AT2G01530
+AT5G65710
+AT1G28815
+AT3G27906
+AT5G04390
+AT4G08410
+AT1G67810
+AT3G54210
+AT5G09520
+AT5G25220
+AT1G33860
+AT5G53742
+AT3G57760
+AT3G57830
+AT5G03100
+AT2G07787
+AT2G03440
+AT2G15300
+AT2G23190
+AT1G64590
+AT5G49080
+AT1G67800
+AT1G27180
+AT3G62725
+ATCG00070
+AT5G45460
+AT4G37000
+AT5G25270
+AT4G28085
+AT5G10230
+AT2G01520
+AT4G12450
+AT3G19380
+AT2G26110
+AT1G77400
+AT1G33100
+AT5G14500
+AT5G40150
+AT2G35910
+AT5G17340
+AT4G14465
+AT1G12330
+AT2G08986
+AT1G72790
+AT1G12060
+AT3G13270
+AT5G26622
+AT2G36470
+AT5G61600
+AT5G06839
+AT2G48000
+AT3G04360
+AT3G52460
+AT4G18335
+AT5G61480
+AT5G24010
+AT5G01830
+AT5G44550
+AT5G55780
+ATCG00640
+AT1G70170
+AT4G12050
+AT5G12420
+AT1G08750
+AT5G14890
+AT5G15830
+AT5G62180
+AT1G69310
+AT2G41870
+AT2G45170
+AT5G28913
+AT1G13540
+AT5G65460
+AT1G19300
+AT5G14570
+AT2G07696
+AT2G07711
+AT3G61460
+AT4G19460
+AT3G55090
+AT5G12990
+AT5G10970
+AT3G54580
+AT1G29290
+AT5G64120
+AT3G57320
+AT5G58680
+AT3G62430
+ATMG00640
+AT2G46590
+AT1G21510
+AT1G05675
+AT5G06640
+AT1G09176
+AT3G18035
+AT3G44430
+ATMG01170
+AT5G07330
+AT5G11990
+AT5G37990
+AT5G22090
+AT3G25940
+AT1G68330
+AT3G49120
+AT3G50280
+AT3G62660
+AT2G44000
+AT5G65080
+AT3G44880
+AT4G24040
+AT1G62440
+AT5G66150
+AT3G11420
+AT3G51740
+AT1G73000
+AT3G01970
+AT4G26320
+AT2G27130
+AT2G17880
+AT1G77525
+AT4G19191
+AT1G49210
+AT2G24980
+AT1G07795
+AT5G19060
+AT2G31110
+AT1G80810
+AT3G14190
+AT5G14140
+AT4G25270
+AT2G30933
+ATMG01380
+AT5G15060
+ATCG00150
+AT5G50530
+AT3G62680
+ATCG00580
+AT5G65170
+AT3G43840
+AT2G47150
+AT4G34135
+AT1G50190
+AT3G11580
+AT1G07590
+AT4G28890
+AT4G09170
+AT5G12020
+AT2G39110
+AT1G08810
+AT4G18090
+AT5G41690
+AT2G07042
+AT2G30230
+AT2G28870
+AT3G01500
+ATCG01070
+AT5G35190
+AT5G51530
+AT3G09032
+AT1G12040
+AT5G64980
+AT5G13790
+AT4G29610
+AT3G45710
+AT1G48745
+AT5G06330
+AT5G61340
+ATCG00430
+AT3G30122
+AT3G28920
+AT4G37540
+AT3G44990
+AT5G65070
+AT5G62170
+AT3G54410
+AT1G43560
+AT4G35900
+AT2G24600
+AT3G14680
+AT4G33990
+AT5G37690
+AT2G35650
+AT3G03550
+AT4G36350
+AT4G03380
+AT3G26210
+AT5G50200
+AT4G39780
+AT5G03180
+AT1G17240
+AT1G03730
+AT3G54680
+AT5G64620
+AT5G38290
+AT5G03050
+AT5G13330
+AT2G35830
+AT1G76100
+AT4G30180
+AT1G80180
+AT2G28305
+AT1G14430
+AT5G25910
+AT5G17600
+AT2G02070
+AT1G12020
+AT1G67090
+AT4G12300
+AT5G38020
+AT4G00070
+AT4G37450
+AT1G80800
+AT5G01910
+AT5G47560
+AT3G12977
+AT2G25260
+AT4G36280
+AT1G69600
+ATCG00350
+AT1G11700
+AT5G49690
+AT3G56270
+AT1G17345
+AT5G17980
+AT3G20640
+AT1G07240
+AT2G40113
+AT2G38380
+AT1G49180
+AT3G43580
+AT2G35750
+AT1G65510
+AT5G28500
+AT5G04470
+AT2G41470
+AT1G63580
+AT5G66320
b
diff -r 000000000000 -r 0a0bba8e1823 test-data/output_newpairs.tsv
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/output_newpairs.tsv Wed Feb 22 07:32:08 2017 -0500
b
b'@@ -0,0 +1,538 @@\n+AT1G76630\tAT1G76640\n+AT2G06045\tAT2G06050\n+AT4G30100\tAT4G30110\n+AT3G63300\tAT3G63310\n+AT1G50740\tAT1G50750\n+AT5G18960\tAT5G18970\n+AT5G07610\tAT5G07620\n+AT1G59860\tAT1G59865\n+AT2G13550\tAT2G13560\n+AT5G07870\tAT5G07880\n+AT3G11810\tAT3G11820\n+AT5G43050\tAT5G43060\n+AT1G32960\tAT1G32970\n+AT1G01730\tAT1G01740\n+AT3G09050\tAT3G09060\n+AT5G67100\tAT5G67110\n+AT1G13930\tAT1G13940\n+AT3G62190\tAT3G62200\n+AT1G16370\tAT1G16380\n+AT3G14395\tAT3G14400\n+AT1G18740\tAT1G18745\n+AT4G28280\tAT4G28290\n+AT4G24570\tAT4G24580\n+AT4G11070\tAT4G11080\n+AT3G23540\tAT3G23550\n+AT5G59190\tAT5G59200\n+AT1G21340\tAT1G21350\n+AT2G40030\tAT2G40050\n+AT2G31090\tAT2G31100\n+AT4G23840\tAT4G23850\n+AT5G52580\tAT5G52600\n+AT2G29430\tAT2G29440\n+AT5G12130\tAT5G12140\n+AT3G62410\tAT3G62420\n+AT5G21930\tAT5G21940\n+AT2G23200\tAT2G23210\n+AT4G03490\tAT4G03500\n+AT4G22930\tAT4G22940\n+AT1G15160\tAT1G15165\n+AT3G15730\tAT3G15740\n+AT4G23270\tAT4G23271\n+AT1G07250\tAT1G07260\n+AT5G53380\tAT5G53390\n+AT3G22900\tAT3G22910\n+AT3G56250\tAT3G56260\n+AT1G15190\tAT1G15200\n+AT1G10160\tAT1G10170\n+AT4G01533\tAT4G01535\n+AT3G61410\tAT3G61415\n+AT4G38080\tAT4G38090\n+AT2G18700\tAT2G18710\n+AT5G02060\tAT5G02070\n+AT1G15260\tAT1G15270\n+AT4G13320\tAT4G13330\n+AT5G05160\tAT5G05170\n+AT4G20420\tAT4G20430\n+AT1G42980\tAT1G42990\n+AT1G74160\tAT1G74170\n+AT4G16480\tAT4G16490\n+AT5G05070\tAT5G05080\n+AT3G18910\tAT3G18930\n+AT3G57130\tAT3G57140\n+AT3G06820\tAT3G06830\n+AT5G56170\tAT5G56180\n+AT5G20250\tAT5G20260\n+AT4G13870\tAT4G13880\n+AT3G16060\tAT3G16070\n+AT2G47090\tAT2G47110\n+AT1G52780\tAT1G52790\n+AT2G28440\tAT2G28450\n+AT1G17500\tAT1G17510\n+AT4G02970\tAT4G02980\n+AT1G06240\tAT1G06250\n+AT5G22810\tAT5G22820\n+AT5G52280\tAT5G52290\n+AT5G17830\tAT5G17840\n+AT4G29870\tAT4G29880\n+AT3G57620\tAT3G57630\n+AT5G19400\tAT5G19410\n+AT5G13280\tAT5G13290\n+AT1G73190\tAT1G73200\n+AT2G17830\tAT2G17840\n+AT2G46020\tAT2G46030\n+AT2G30690\tAT2G30695\n+AT3G24929\tAT3G24982\n+AT3G51540\tAT3G51550\n+AT1G03990\tAT1G04000\n+AT4G37710\tAT4G37720\n+AT5G49060\tAT5G49070\n+AT3G52080\tAT3G52090\n+AT5G04660\tAT5G04670\n+AT3G19240\tAT3G19250\n+AT2G33060\tAT2G33070\n+AT5G10600\tAT5G10605\n+AT3G05310\tAT3G05320\n+AT5G10340\tAT5G10350\n+AT5G66030\tAT5G66040\n+AT3G12810\tAT3G12820\n+AT5G04720\tAT5G04730\n+AT3G02380\tAT3G02390\n+AT1G26700\tAT1G26710\n+AT1G71740\tAT1G71750\n+AT1G72710\tAT1G72720\n+AT4G32810\tAT4G32820\n+AT1G70100\tAT1G70110\n+AT1G34420\tAT1G34430\n+AT1G71110\tAT1G71120\n+AT5G64030\tAT5G64040\n+AT5G01670\tAT5G01680\n+AT3G10820\tAT3G10840\n+AT3G18480\tAT3G18485\n+AT3G52250\tAT3G52260\n+AT3G17920\tAT3G17930\n+AT1G05510\tAT1G05520\n+AT1G16000\tAT1G16010\n+AT1G77370\tAT1G77380\n+AT1G16250\tAT1G16260\n+AT5G55530\tAT5G55540\n+AT2G43480\tAT2G43490\n+AT1G07850\tAT1G07870\n+AT2G31820\tAT2G31830\n+AT2G40850\tAT2G40860\n+AT4G37690\tAT4G37700\n+AT1G34100\tAT1G34110\n+AT3G03360\tAT3G03370\n+AT2G46735\tAT2G46740\n+AT4G19210\tAT4G19220\n+AT1G33612\tAT1G33615\n+AT2G24960\tAT2G24970\n+AT1G35720\tAT1G35730\n+AT2G40580\tAT2G40590\n+AT1G79950\tAT1G79960\n+AT4G25860\tAT4G25870\n+AT4G38590\tAT4G38600\n+AT5G64270\tAT5G64280\n+AT1G14190\tAT1G14200\n+AT2G22870\tAT2G22880\n+AT5G02150\tAT5G02160\n+AT1G68862\tAT1G68870\n+AT1G48598\tAT1G48605\n+AT1G75261\tAT1G75270\n+AT4G31740\tAT4G31750\n+AT5G42290\tAT5G42300\n+AT3G05630\tAT3G05640\n+AT5G14160\tAT5G14170\n+AT4G32830\tAT4G32840\n+AT3G62970\tAT3G62980\n+AT1G30620\tAT1G30630\n+AT5G13580\tAT5G13590\n+AT3G59500\tAT3G59510\n+AT4G23000\tAT4G23010\n+AT1G51620\tAT1G51630\n+AT2G37690\tAT2G37700\n+AT1G54490\tAT1G54500\n+AT2G41820\tAT2G41830\n+AT1G73885\tAT1G73890\n+AT1G17200\tAT1G17210\n+AT3G04630\tAT3G04640\n+AT5G65510\tAT5G65520\n+AT1G52820\tAT1G52825\n+AT4G22520\tAT4G22530\n+AT4G29580\tAT4G29590\n+AT4G14890\tAT4G14900\n+AT1G26260\tAT1G26270\n+AT2G16390\tAT2G16400\n+AT3G16460\tAT3G16470\n+AT4G36540\tAT4G36550\n+AT2G40160\tAT2G40170\n+AT1G27000\tAT1G27008\n+AT5G06930\tAT5G06940\n+AT1G15570\tAT1G15580\n+AT1G71400\tAT1G71410\n+AT1G05630\tAT1G05640\n+AT1G22660\tAT1G22670\n+AT1G30455\tAT1G30460\n+AT4G39400\tAT4G39403\n+AT4G39220\tAT4G39230\n+AT5G05410\tAT5G05420\n+AT1G24159\tAT1G24160\n+AT2G34440\tAT2G34450\n+AT5G41910\tAT5G41920\n+AT5G60170\tAT5G60180\n+AT5G03300\tAT5G03310\n+AT1G53190\tAT1G53200\n+AT2G46570\tAT2G46580\n+AT2G01110\tAT2G01120\n+AT5G17500\tAT5G17510\n+AT1G11060\tAT1G11070\n+AT2G28160\tAT2G28170\n+AT5G46090\tA'..b'T5G12980\n+AT4G16110\tAT4G16120\n+AT1G31910\tAT1G31920\n+AT1G09410\tAT1G09415\n+AT5G66070\tAT5G66080\n+AT2G48130\tAT2G48140\n+AT4G33600\tAT4G33610\n+AT2G47820\tAT2G47830\n+AT1G28680\tAT1G28690\n+AT1G65840\tAT1G65845\n+AT5G14090\tAT5G14100\n+AT3G44180\tAT3G44190\n+ATCG01040\tATCG01050\n+AT2G24681\tAT2G24683\n+AT2G34260\tAT2G34270\n+AT2G16630\tAT2G16640\n+AT5G34930\tAT5G34940\n+AT2G31910\tAT2G31920\n+AT5G21070\tAT5G21080\n+AT2G07681\tAT2G07682\n+AT5G28900\tAT5G28910\n+AT1G17470\tAT1G17480\n+AT4G01990\tAT4G01995\n+AT4G11330\tAT4G11340\n+AT3G20280\tAT3G20290\n+AT2G25560\tAT2G25565\n+AT5G20020\tAT5G20030\n+AT2G01710\tAT2G01720\n+AT2G47920\tAT2G47930\n+AT1G10030\tAT1G10040\n+AT2G42990\tAT2G43000\n+AT5G67385\tAT5G67390\n+AT4G11690\tAT4G11700\n+AT2G03380\tAT2G03390\n+AT4G03410\tAT4G03415\n+AT4G38340\tAT4G38350\n+AT4G37480\tAT4G37490\n+AT4G20890\tAT4G20900\n+AT1G23010\tAT1G23020\n+AT2G32640\tAT2G32650\n+AT1G29380\tAT1G29390\n+AT4G25250\tAT4G25260\n+AT4G14520\tAT4G14530\n+ATCG00650\tATCG00660\n+AT3G19220\tAT3G19230\n+AT4G31805\tAT4G31810\n+AT1G68020\tAT1G68030\n+AT4G33060\tAT4G33070\n+AT5G45430\tAT5G45440\n+AT4G14280\tAT4G14290\n+AT5G46030\tAT5G46040\n+AT1G07705\tAT1G07710\n+AT4G10640\tAT4G10650\n+AT2G39190\tAT2G39200\n+AT5G10605\tAT5G10610\n+AT5G10050\tAT5G10060\n+AT1G61520\tAT1G61665\n+AT4G19080\tAT4G19090\n+AT2G17140\tAT2G17150\n+AT5G18650\tAT5G18660\n+AT5G09220\tAT5G09225\n+AT1G32850\tAT1G32860\n+AT5G18890\tAT5G18900\n+AT1G09070\tAT1G09080\n+AT1G78300\tAT1G78310\n+AT1G16060\tAT1G16070\n+AT5G14360\tAT5G14370\n+AT2G33570\tAT2G33580\n+AT2G42720\tAT2G42730\n+AT5G02080\tAT5G02090\n+AT3G01890\tAT3G01900\n+AT3G09370\tAT3G09375\n+AT4G39270\tAT4G39280\n+AT1G79400\tAT1G79410\n+AT2G45310\tAT2G45320\n+AT5G61030\tAT5G61040\n+AT2G07697\tAT2G07798\n+AT1G78500\tAT1G78510\n+AT1G14390\tAT1G14400\n+AT4G14180\tAT4G14190\n+AT5G48620\tAT5G48630\n+AT3G61700\tAT3G61710\n+AT1G73340\tAT1G73350\n+AT5G07640\tAT5G07650\n+AT4G30130\tAT4G30140\n+AT1G11410\tAT1G11420\n+AT5G48920\tAT5G48930\n+AT3G15430\tAT3G15440\n+AT2G17440\tAT2G17442\n+AT3G60286\tAT3G60290\n+AT3G15110\tAT3G15111\n+AT5G40440\tAT5G40450\n+AT1G77290\tAT1G77300\n+AT2G42620\tAT2G42640\n+AT4G13290\tAT4G13300\n+AT1G45229\tAT1G45233\n+AT1G63300\tAT1G63310\n+AT1G33770\tAT1G33780\n+AT5G24460\tAT5G24470\n+AT1G09560\tAT1G09570\n+AT3G48580\tAT3G48590\n+AT1G23790\tAT1G23800\n+AT3G58570\tAT3G58580\n+AT3G48160\tAT3G48170\n+AT2G44010\tAT2G44020\n+AT4G33160\tAT4G33170\n+AT1G06030\tAT1G06040\n+AT3G02160\tAT3G02170\n+AT4G00895\tAT4G00900\n+AT5G25630\tAT5G25640\n+AT5G66830\tAT5G66840\n+AT5G15330\tAT5G15340\n+AT5G04200\tAT5G04210\n+AT1G69010\tAT1G69020\n+AT2G40935\tAT2G40940\n+AT1G03810\tAT1G03820\n+AT4G02655\tAT4G02660\n+AT5G10680\tAT5G10690\n+AT2G22080\tAT2G22090\n+AT4G14548\tAT4G14550\n+AT2G29640\tAT2G29650\n+AT1G02700\tAT1G02710\n+AT1G48920\tAT1G48930\n+AT4G19540\tAT4G19550\n+AT4G23930\tAT4G23940\n+AT5G67470\tAT5G67480\n+AT1G74680\tAT1G74690\n+AT3G11460\tAT3G11470\n+AT1G05140\tAT1G05150\n+AT1G33470\tAT1G33475\n+AT3G12760\tAT3G12770\n+AT4G33925\tAT4G33930\n+AT1G49890\tAT1G49900\n+AT3G50230\tAT3G50240\n+AT3G10310\tAT3G10320\n+AT5G25310\tAT5G25320\n+AT5G43740\tAT5G43745\n+AT5G08020\tAT5G08030\n+AT1G08600\tAT1G08610\n+AT2G33450\tAT2G33460\n+AT3G25495\tAT3G25500\n+AT3G12060\tAT3G12070\n+AT1G67430\tAT1G67440\n+AT4G31870\tAT4G31875\n+AT3G43260\tAT3G43270\n+AT3G07900\tAT3G07910\n+AT3G18850\tAT3G18860\n+AT2G30780\tAT2G30790\n+AT2G32960\tAT2G32970\n+AT5G09225\tAT5G09230\n+AT1G68110\tAT1G68120\n+AT1G72660\tAT1G72670\n+AT3G50910\tAT3G50920\n+AT3G42660\tAT3G42670\n+AT5G52170\tAT5G52180\n+AT3G16980\tAT3G16990\n+AT5G58375\tAT5G58380\n+AT5G56870\tAT5G56880\n+AT3G23480\tAT3G23490\n+AT5G03545\tAT5G03550\n+AT5G10750\tAT5G10760\n+AT1G02010\tAT1G02020\n+AT2G04350\tAT2G04360\n+AT3G14410\tAT3G14415\n+AT3G12360\tAT3G12370\n+AT1G53270\tAT1G53280\n+AT4G16710\tAT4G16720\n+AT1G77530\tAT1G77540\n+AT2G42750\tAT2G42760\n+AT1G03770\tAT1G03780\n+AT2G01735\tAT2G01740\n+ATCG00690\tATCG00700\n+AT2G26340\tAT2G26350\n+AT2G45440\tAT2G45450\n+AT2G21550\tAT2G21560\n+AT1G29520\tAT1G29530\n+AT1G14270\tAT1G14280\n+AT4G03120\tAT4G03130\n+AT3G14580\tAT3G14590\n+AT2G44735\tAT2G44740\n+AT1G80940\tAT1G80950\n+AT2G27505\tAT2G27510\n+AT3G22945\tAT3G22950\n+AT2G41750\tAT2G41760\n+AT3G12080\tAT3G12090\n+AT4G21700\tAT4G21705\n+AT2G02760\tAT2G02770\n+AT4G13860\tAT4G13870\n+AT1G20640\tAT1G20650\n+AT3G01280\tAT3G01290\n+AT5G57340\tAT5G57345\n'
b
diff -r 000000000000 -r 0a0bba8e1823 test-data/test.r
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/test.r Wed Feb 22 07:32:08 2017 -0500
[
@@ -0,0 +1,53 @@
+library(NASTIseq)
+
+## generation of test set
+# data(WholeRoot)
+# WholeRoot$genepos$feature <- 'gene'
+#
+# set_attri <- function(attri){
+#   attri = paste('gene_id ', '"', attri, '"', ';', sep = '')
+#   return(attri)
+# }
+#
+# WholeRoot$genepos$attributes = as.character(lapply(as.character(WholeRoot$genepos$attributes), set_attri))
+#
+# write.table(WholeRoot$genepos, file = "input_TAIR10_annotation.gtf", row.names = FALSE,  col.names = FALSE,  sep = "\t",  quote = FALSE)
+# write.table(WholeRoot$smat, file = "input_read_count_smt.tsv",  col.names = FALSE,  sep = "\t", quote = FALSE)
+# write.table(WholeRoot$asmat, file = "input_read_count_asmt.tsv", col.names = FALSE, sep = "\t", quote = FALSE)
+#
+# write.table(WholeRoot$pospairs, file = "input_positive_pair.tsv", row.names = FALSE, col.names = FALSE, sep = "\t", quote = FALSE)
+
+genepos = read.delim("input_TAIR10_annotation.gtf", header=FALSE, comment.char="#")
+colnames(genepos) = c("seqname", "source", "feature", "start", "end", "score", "strand", "frame", "attributes")
+genepos = subset(genepos, feature=="gene")
+
+get_id = function(attri){
+  gene_info = strsplit(attri, ";")[[1]][1]
+  gene_id = strsplit(gene_info, " ")[[1]][2]
+  gene_id = gsub("\"", "", gene_id)
+  return(gene_id)
+}
+
+genepos$attributes = as.character(lapply(as.character(genepos$attributes), get_id))
+
+pospairs = read.table("input_positive_pair.tsv", sep = "\t", as.is = TRUE)
+
+smat = as.matrix(read.table("input_read_count_smt.tsv",  sep = "\t",  row.names = 1))
+
+asmat = as.matrix(read.table("input_read_count_asmt.tsv",  sep = "\t",  row.names = 1))
+
+WRscore = getNASTIscore(smat, asmat)
+
+negpairs = getnegativepairs(genepos)
+
+WRpred = NASTIpredict(smat,asmat, pospairs, negpairs)
+
+WRpred_rocr = prediction(WRpred$predictions,WRpred$labels)
+
+thr = defineFDR(WRpred_rocr,0.05)
+
+WR_names = FindNATs(WRscore, thr, pospairs, genepos)
+
+write.table(WR_names$newpairs, file = "output_newpairs.tsv", row.names = FALSE, col.names = FALSE, sep = "\t", quote = FALSE)
+
+write.table(WR_names$neworphan, file = "output_neworphan.tsv", row.names = FALSE, col.names = FALSE, sep = "\t", quote = FALSE)