From 2648968c549e5b67d383b2af5d76fd3a1a9b9977 Mon Sep 17 00:00:00 2001
From: Heng Li <lh3@me.com>
Date: Tue, 24 Oct 2023 21:29:05 -0400
Subject: [PATCH] Release pangene-1.0 (r183)

---
 cmd-log.sh | 17 +++++++++++++++++
 pangene.1  |  2 +-
 pangene.h  |  2 +-
 3 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/cmd-log.sh b/cmd-log.sh
index a73963e..4bcf66e 100644
--- a/cmd-log.sh
+++ b/cmd-log.sh
@@ -37,3 +37,20 @@ pangene -P Mtb.H37Rv.txt -p.001 aln/*.paf.gz 2> Mtb-merge-p0.r177.gfa.log | gzip
 # run panaroo
 runlog panaroo -i *.gff -o panaroo.strict -t 16 --clean-mode strict > panaroo.strict.log 2>&1
 runlog panaroo -i *.gff -o panaroo.strict-merge -t 16 --clean-mode strict --merge_paralogs > panaroo.strict-merge.log 2>&1
+
+#############
+### Human ###
+#############
+
+ls *.fa.gz | sed s,.fa.gz,, | xargs -i echo miniprot --outs=0.97 --no-cs -Iut16 {}.fa.gz ../HUMAN.faa.gz \| gzip \> {}.paf.gz|asub -j run-mp -M36 -n16
+../pangene -a2 -I@gape.huamn.incl -X@gape.human.excl aln-human/[14]*.paf.gz 2> human-human98-r182.gfa.log | gzip > human-human98-r182.gfa.gz
+
+##################
+### Great apes ###
+##################
+
+# generate proteins
+cat HUMAN.faa.gz PANTR.faa.gz PANPA.faa.gz GROGO.faa.gz PONAB.faa.gz > merged.faa.gz
+ls *.fa.gz|sed s,.fa.gz,,|xargs -i echo miniprot --outs=0.97 --no-cs -Iut16 {}.fa.gz merged.faa.gz \| gzip \> {}.paf.gz | asub -j run-mp -M40 -n16
+../pangene 1100_GRCh38.0.paf.gz 7031_chimpanzee.paf.gz 7032_bonobo.paf.gz 7033_gorilla.paf.gz 7034_orangutan.paf.gz > 5a.gfa
+(zcat ../proteins/HUMAN.faa.gz; seqtk comp merged.faa.gz|cut -f1|grep -wFf <(../pangene.js gfa2matrix 5a.gfa|grep _|awk '$2==0'|cut -f1) | seqtk subseq -l80 merged.faa.gz -) | gzip > human+other.faa.gz
diff --git a/pangene.1 b/pangene.1
index b213233..f775f3d 100644
--- a/pangene.1
+++ b/pangene.1
@@ -1,4 +1,4 @@
-.TH pangene 1 "19 October 2023" "pangene-0.0-dirty (r177)" "Bioinformatics tools"
+.TH pangene 1 "24 October 2023" "pangene-1.0 (r183)" "Bioinformatics tools"
 .SH NAME
 .PP
 pangene - building pangenome gene graphs
diff --git a/pangene.h b/pangene.h
index 0f8217f..2affc3c 100644
--- a/pangene.h
+++ b/pangene.h
@@ -3,7 +3,7 @@
 
 #include <stdint.h>
 
-#define PG_VERSION "0.0-r182-dirty"
+#define PG_VERSION "1.0-r183-dirty"
 
 #define PG_F_WRITE_BED_RAW      0x1
 #define PG_F_WRITE_BED_WALK     0x2