From 582508dbc5a44daa0498d643e6163413df072832 Mon Sep 17 00:00:00 2001
From: Arun Somwarpet-Seetharam - staff <arnstrm@nova.its.iastate.edu>
Date: Sun, 24 Feb 2019 19:38:48 -0600
Subject: [PATCH 1/2] script for creating ini file for i-AdHore, using just the
 gff files

---
 generate_i-AdHore_ini-file.sh | 47 +++++++++++++++++++++++++++++++++++
 1 file changed, 47 insertions(+)
 create mode 100755 generate_i-AdHore_ini-file.sh

diff --git a/generate_i-AdHore_ini-file.sh b/generate_i-AdHore_ini-file.sh
new file mode 100755
index 0000000..7fbb981
--- /dev/null
+++ b/generate_i-AdHore_ini-file.sh
@@ -0,0 +1,47 @@
+#!/bin/bash
+gff="$1"
+base="$2"
+#table="$3"
+mkdir -p $base
+awk '$3=="mRNA" {print $1"\t"$7"\t"$9}' $gff |\
+     sed 's/;/\t/1' | sed 's/ID=//g' |\
+     awk -v x="$base" '{print x"_"$1"\t"$3""$2}' |\
+     sed 's/Sobic./'$base'_/g' >  $base/${base}_full.lst
+awk '$3=="mRNA" {print $9}' $gff |\
+     sed -e 's/;/\t/1' -e 's/ID=//g' |\
+     cut -f 1  |\
+     awk '{print $1"\tfamily."$1}' |\
+     sed -e 's/family.Sobic./OG_/g' -e 's/.mrna.\{1,3\}$//g' |\
+     sed 's/^Sobic./'$base'_/1'  >>  blast_table.txt
+
+awk '{print>$1"_chr.list"}'  ${base}/${base}_full.lst
+#echo "blast_table=$(realpath ${base}_blast.pairs)"
+echo "genome=${base}"
+i=1
+for list in ${base}*_chr.list; do
+cut -f 2 $list > $list.temp;
+mv $list.temp $base/$list;
+echo "CO$i ${base}/${list}"
+rm $list
+((i++))
+done
+cat <<EOF
+blast_table=blast_table.txt
+table_type=family
+cluster_type=colinear
+tandem_gap=10
+prob_cutoff=0.1
+write_stats=true
+multiple_hypothesis_correction=FDR
+gap_size=15
+cluster_gap=20
+q_value=0.7
+anchor_points=5
+alignment_method=gg2
+max_gaps_in_alignment=20
+output_path=$(pwd)/iadhore_output2
+number_of_threads=16
+visualizeGHM=true
+visualizeAlignment=true
+verbose_output=true
+EOF

From 1537ce56d157e6f6cf77641499e1c4a699ddf82d Mon Sep 17 00:00:00 2001
From: Arun Seetharam <arnstrm@iastate.edu>
Date: Mon, 25 Feb 2019 09:21:36 -0600
Subject: [PATCH 2/2] script for aligning cds to genome to get the gff3 file

---
 runGMAP.sh | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
 create mode 100755 runGMAP.sh

diff --git a/runGMAP.sh b/runGMAP.sh
new file mode 100755
index 0000000..5f4e89e
--- /dev/null
+++ b/runGMAP.sh
@@ -0,0 +1,14 @@
+#!/bin/bash
+module load gmap-gsnap
+db="$1"
+fasta="$2"
+threads=36
+out="${db%.*}_${fasta%.*}"
+for type in match_est match_cdna gene; do
+if [ ! -f "${out}_${type}.gff3.done" ]; then
+gmap -D /work/LAS/mhufford-lab/arnstrm/NAM/genomes -d ${db} -B 4 -t $threads -f gff3_${type} ${fasta} > ${out}_${type}.gff3 2> ${out}_${type}.err
+if [ $? -eq 0 ]; then
+    touch "${out}_${type}.gff3.done"
+fi
+fi
+done