From 1b73c7cefd9730c121cabfb39d48300789b4f889 Mon Sep 17 00:00:00 2001
From: Glenn Hickey <glenn.hickey@gmail.com>
Date: Thu, 16 Mar 2023 11:40:34 -0400
Subject: [PATCH] add temp N filter for maf reflines in hal2maf

---
 build-tools/downloadMafTools     | 2 +-
 doc/progressive.md               | 4 +++-
 src/cactus/maf/cactus_hal2maf.py | 9 ++++++++-
 3 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/build-tools/downloadMafTools b/build-tools/downloadMafTools
index e6e83f101..ac30273b3 100755
--- a/build-tools/downloadMafTools
+++ b/build-tools/downloadMafTools
@@ -55,7 +55,7 @@ fi
 cd ${mafBuildDir}
 git clone https://github.com/ComparativeGenomicsToolkit/mafTools.git
 cd mafTools
-git checkout 40cfa5b503a34b8b0b7799678237e2f13ae8bf36
+git checkout 0d2a253a528749bad2c6c0179bd15edd8d56adf6
 find . -name "*.mk" | xargs sed -ie "s/-Werror//g"
 find . -name "Makefile*" | xargs sed -ie "s/-Werror//g"
 # hack in flags support
diff --git a/doc/progressive.md b/doc/progressive.md
index 9be688c98..b4202b44c 100644
--- a/doc/progressive.md
+++ b/doc/progressive.md
@@ -143,7 +143,9 @@ cactus-hal2chain ./js ./evolverMammals.hal chains-dir --refGenome simHuman_chr6
 
 will create `./chains-dir` and populate it with a Chain alignment between simHuman and each other leaf genome in evolverMammals.hal.
 
-By default, chains will be created using `halLiftover` [as in CAT](https://github.com/ComparativeGenomicsToolkit/Comparative-Annotation-Toolkit/blob/fc1623da5df1309d2e2f0b9bb0363aaab84708f4/cat/chaining.py#L96-L98). An option `--useHalSynteny` is provided to use that tool instead. 
+By default, chains will be created using `halLiftover` [as in CAT](https://github.com/ComparativeGenomicsToolkit/Comparative-Annotation-Toolkit/blob/fc1623da5df1309d2e2f0b9bb0363aaab84708f4/cat/chaining.py#L96-L98). An option `--useHalSynteny` is provided to use that tool instead.
+
+See here for an all-vs-all script to make chains, including BigChain conversion: https://github.com/human-pangenomics/HPRC_Assembly_Hub/blob/main/chains/wdl/snakesonachain.wdl
 
 ### CAT
 
diff --git a/src/cactus/maf/cactus_hal2maf.py b/src/cactus/maf/cactus_hal2maf.py
index 636b2f4dc..b1d37c117 100644
--- a/src/cactus/maf/cactus_hal2maf.py
+++ b/src/cactus/maf/cactus_hal2maf.py
@@ -98,7 +98,12 @@ def main():
     parser.add_argument("--keepGapCausingDupes",
                         help="Turn off taffy norm -d filter that removes duplications that would induce gaps > maximumGapLength",
                         action="store_true")
-    
+
+    parser.add_argument("--maxRefNFrac",
+                        help="(hopefully temporary) partial work around of a current bug that aligns through Ns by filtering out MAF blocks whose reference (first) line has a greater fraction of Ns than the given amount. Should be between 0.0 (filter everything) and 1.0 (filter nothing). [default=0.75]",
+                        type=float,
+                        default=0.75)
+                        
     #Progressive Cactus Options
     parser.add_argument("--configFile", dest="configFile",
                         help="Specify cactus configuration file",
@@ -301,6 +306,8 @@ def taf_cmd(hal_path, chunk, chunk_num, options):
     cmd += ' | {} taffy norm -k -m {} -n {} {} -q {}{} 2> {}.tn.time'.format(time_cmd, options.maximumBlockLengthToMerge, options.maximumGapLength,
                                                                              '' if options.keepGapCausingDupes else '-d',
                                                                              options.fractionSharedRows, time_end, chunk_num)
+    if options.maxRefNFrac:
+        cmd += ' | mafFilter -m - -N {}'.format(options.maxRefNFrac)
     if options.dupeMode == 'single':
         cmd += ' | mafDuplicateFilter -m - -k'
     if chunk[1] != 0: