From ed98174bbefc85f9895c870fdb7518e2a534a150 Mon Sep 17 00:00:00 2001
From: Justin Bedo <cu@cua0.org>
Date: Wed, 28 Jul 2021 08:07:32 +1000
Subject: bugfix: use unsorted type for alignment output

Most aligners are only pseudo-name sorted (i.e., pairs adjacant).
---
 examples/call.nix             | 1 +
 examples/ex-tnpair/tnpair.nix | 1 +
 test-tnpair.nix               | 4 ++--
 tools/bowtie-align.nix        | 2 +-
 tools/bwa-mem.nix             | 2 +-
 tools/bwa-mem2.nix            | 2 +-
 tools/gridss-assemble.nix     | 2 +-
 tools/hisat2-align.nix        | 2 +-
 tools/minimap2-align.nix      | 2 +-
 tools/snap-align.nix          | 2 +-
 tools/star-align.nix          | 2 +-
 tools/whisper-align.nix       | 2 +-
 12 files changed, 13 insertions(+), 11 deletions(-)

diff --git a/examples/call.nix b/examples/call.nix
index 8c4f073..8a38fdb 100644
--- a/examples/call.nix
+++ b/examples/call.nix
@@ -12,6 +12,7 @@ with lib;
 let
   preprocess = flip pipe [
     (bwa.align { inherit ref; })
+    (samtools.sort { nameSort = true; })
     (samtools.fixmate {})
     (samtools.sort {})
     (samtools.markdup {})
diff --git a/examples/ex-tnpair/tnpair.nix b/examples/ex-tnpair/tnpair.nix
index 70f5514..9e6befe 100644
--- a/examples/ex-tnpair/tnpair.nix
+++ b/examples/ex-tnpair/tnpair.nix
@@ -12,6 +12,7 @@ let
   preprocess = s: pipe s [
     fetch
     (align { preset = "sr"; ref = ref.grch38.seq; flags = "-R'@RG\\tID:${s.type}\\tSM:${s.type}'"; })
+    (sort { nameSort = true; })
     (fixmate {})
     (sort { })
     (markdup { })
diff --git a/test-tnpair.nix b/test-tnpair.nix
index 7cc5453..6eb2e23 100644
--- a/test-tnpair.nix
+++ b/test-tnpair.nix
@@ -11,7 +11,7 @@ let
 
   ref = fetchfa ./examples/ref.fa;
 
-  alignWithRG = rg: bwa.align { inherit ref; flags = "-R'@RG\\tID:${rg}\\tSM:${rg}'";};
+  alignWithRG = rg: x: nameSort (bwa.align { inherit ref; flags = "-R'@RG\\tID:${rg}\\tSM:${rg}'";} x);
   sort = samtools.sort {};
   nameSort = samtools.sort {nameSort = true;};
   flagstat = samtools.flagstat {};
@@ -41,7 +41,7 @@ let
     glvariants = strelka.call {} (builtins.attrValues alignments);
     platypusVars = platypus.call {} (builtins.attrValues alignments);
     octopusVars = octopus.call {} (builtins.attrValues alignments);
-    shards = map (bwa.align {inherit ref;}) (shard.fastQPair 2 normal.files);
+    shards = map (x: nameSort (bwa.align {inherit ref;} x)) (shard.fastQPair 2 normal.files);
   };
 
   tnpairResult = processPair tnpair;
diff --git a/tools/bowtie-align.nix b/tools/bowtie-align.nix
index 57b0b01..c83c49c 100644
--- a/tools/bowtie-align.nix
+++ b/tools/bowtie-align.nix
@@ -32,6 +32,6 @@ in stage {
       ${optionalString bamOutput "| samtools view -b"} \
       > $out
   '';
-  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.name {};} else filetype.sam {ref = ref; sorting = sort.name {};};
+  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};};
   passthru.multicore = true;
 }
diff --git a/tools/bwa-mem.nix b/tools/bwa-mem.nix
index e58ef03..31c9487 100644
--- a/tools/bwa-mem.nix
+++ b/tools/bwa-mem.nix
@@ -35,6 +35,6 @@ in stage {
       ${optionalString bamOutput "| samtools view -b"} \
       > $out
   '';
-  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.name {};} else filetype.sam {ref = ref; sorting = sort.name {};};
+  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};};
   passthru.multicore = true;
 }
diff --git a/tools/bwa-mem2.nix b/tools/bwa-mem2.nix
index 3de1944..b16b7ea 100644
--- a/tools/bwa-mem2.nix
+++ b/tools/bwa-mem2.nix
@@ -35,6 +35,6 @@ in stage {
       ${optionalString bamOutput "| samtools view -b"} \
       > $out
   '';
-  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.name {};} else filetype.sam {ref = ref; sorting = sort.name {};};
+  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};};
   passthru.multicore = true;
 }
diff --git a/tools/gridss-assemble.nix b/tools/gridss-assemble.nix
index 9e51cab..1a46eae 100644
--- a/tools/gridss-assemble.nix
+++ b/tools/gridss-assemble.nix
@@ -70,6 +70,6 @@ stage rec {
     touch $out
     cp -r out.gridss.working $work
   '';
-  passthru.filetype = filetype.bam { ref = ref; sorting = sort.name {}; };
+  passthru.filetype = filetype.bam { ref = ref; sorting = sort.none {}; };
   passthru.multicore = true;
 }
diff --git a/tools/hisat2-align.nix b/tools/hisat2-align.nix
index dfd9601..7a216a6 100644
--- a/tools/hisat2-align.nix
+++ b/tools/hisat2-align.nix
@@ -33,6 +33,6 @@ in stage {
       | samtools sort -n \
       > $out
   '';
-  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.name {};} else filetype.sam {ref = ref; sorting = sort.name {};};
+  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};};
   passthru.multicore = true;
 }
diff --git a/tools/minimap2-align.nix b/tools/minimap2-align.nix
index 7164ace..a6de0b7 100644
--- a/tools/minimap2-align.nix
+++ b/tools/minimap2-align.nix
@@ -32,6 +32,6 @@ in stage {
       ${optionalString bamOutput "| samtools view -b"} \
       > $out
   '';
-  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.name {};} else filetype.sam {ref = ref; sorting = sort.name {};};
+  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.none {};};
   passthru.multicore = true;
 }
diff --git a/tools/snap-align.nix b/tools/snap-align.nix
index 4647142..279cf20 100644
--- a/tools/snap-align.nix
+++ b/tools/snap-align.nix
@@ -32,6 +32,6 @@ in stage {
       ${optionalString (flags != null) flags} \
       | samtools sort -n > $out
   '';
-  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.name {};} else filetype.sam {ref = ref; sorting = sort.name {};};
+  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};};
   passthru.multicore = true;
 }
diff --git a/tools/star-align.nix b/tools/star-align.nix
index 0cf9ab9..44dd7f2 100644
--- a/tools/star-align.nix
+++ b/tools/star-align.nix
@@ -33,6 +33,6 @@ in stage {
      --readFilesIn ${fq input1} ${optionalString (input2 != null) (fq input2)}
      ${if bamOutput then "samtools view -b Aligned.out.sam > $out" else "cp Aligned.out.sam $out"}
   '';
-  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.name {};} else filetype.sam {ref = ref; sorting = sort.name {};};
+  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};};
   passthru.multicore = true;
 }
diff --git a/tools/whisper-align.nix b/tools/whisper-align.nix
index c44205a..1731bfe 100644
--- a/tools/whisper-align.nix
+++ b/tools/whisper-align.nix
@@ -34,6 +34,6 @@ in stage {
       ${optionalString (input2 != null) (fq input2)} \
       > $out
   '';
-  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.name {};} else filetype.sam {ref = ref; sorting = sort.name {};};
+  passthru.filetype = if bamOutput then filetype.bam {ref = ref; sorting = sort.none {};} else filetype.sam {ref = ref; sorting = sort.name {};};
   passthru.multicore = true;
 }
-- 
cgit v1.2.3