From d008e10eec3ee484217582c26af63a10093ad468 Mon Sep 17 00:00:00 2001
From: "Brian P. Walenz" <thebri@gmail.com>
Date: Mon, 21 Sep 2020 12:54:43 -0400
Subject: [PATCH] Script for generating consensus from bogart ctgStore output.

---
 src/pipelines/layout-to-consensus.sh | 44 ++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)
 create mode 100644 src/pipelines/layout-to-consensus.sh

diff --git a/src/pipelines/layout-to-consensus.sh b/src/pipelines/layout-to-consensus.sh
new file mode 100644
index 000000000..7d8cd5389
--- /dev/null
+++ b/src/pipelines/layout-to-consensus.sh
@@ -0,0 +1,44 @@
+#!/bin/sh
+
+#  Runs consensus, in parallel, on a ctgStore from bogart.
+#
+#  Magic values for partitioning:
+#    partitionSize    - make partitions this big, relative to the biggest contig
+#                       0.10 - each big contig gets its own partition
+#    partitionScaling - expect contigs to expand by this much due to homopoly compression
+#                       1.50 - 50% expansion of contig length, affects memory estimate
+#    partitionTigs    - put this many reads (as fraction of total) in a partition
+#                       0.01 - puts small contigs into many partitions
+
+asmpre=asm
+cnspre=asm
+errorRate=0.01
+
+if [ ! -e ./${cnspre}.ctgStore/partitioning ] ; then
+  echo Partitioning reads.
+
+  utgcns -V \
+    -S ../../${asmpre}.seqStore \
+    -T ./${cnspre}.ctgStore 1 \
+    -partition 0.10 1.50 0.01 \
+  > ./${cnspre}.ctgStore/partitioning.log 2>&1
+fi
+
+parts=`cd ./${cnspre}.ctgStore ; ls partition.???? | sed s/partition.//`
+
+for pp in $parts ; do
+  if [ ! -e "./${cnspre}.${pp}.cns" ] ; then
+    echo \
+    utgcns \
+      -R ./${cnspre}.ctgStore/partition.${pp} \
+      -T ./${cnspre}.ctgStore 1 \
+      -P ${pp} \
+      -A ./${cnspre}.${pp}.cns.fasta \
+      -O ./${cnspre}.${pp}.cns \
+      -maxcoverage 50 \
+      -e ${errorRate} \
+      -threads 8 \&
+  fi
+done
+
+exit 0