9ae312b796eca29cd110e08ecc8506ca593fcbfd
angie
  Tue Jun 3 11:22:47 2025 -0700
Relax branch length filter again (for BA.3.2).

diff --git src/hg/utils/otto/sarscov2phylo/updateCombinedTree.sh src/hg/utils/otto/sarscov2phylo/updateCombinedTree.sh
index cc74e885143..3cfa1cbde34 100755
--- src/hg/utils/otto/sarscov2phylo/updateCombinedTree.sh
+++ src/hg/utils/otto/sarscov2phylo/updateCombinedTree.sh
@@ -48,57 +48,57 @@
     gunzip -f new.masked.mpl.gz
     time $usher \
         -T 50 -A -e 5 \
         -i prevRenamed.pb.gz \
         --diff new.masked.mpl \
         --ref $ottoDir/NC_045512.2.fa \
         -o merged.pb.gz \
         --optimization_radius 0 --batch_size_per_process 10 \
         > usher.addNew.log 2>usher-sampled.stderr
     pigz -f -p 8 new.masked.mpl
     # Branch-specific masking
     time $scriptDir/maskDelta.sh merged.pb.gz merged.deltaMasked.pb.gz
     # Prune samples with too many private mutations and internal branches that are too long.
     $matUtils extract -i merged.deltaMasked.pb.gz \
         --max-parsimony 20 \
-        --max-branch-length 60 \
+        --max-branch-length 70 \
         --max-path-length 225 \
         -O -o merged.deltaMasked.filtered.pb.gz
     # matOptimize: used -r 8 -M2 until 2023-05-12, then switched to Cheng's recommended
     # -m 0.00000001 -M 4 (avoid identical-child-node problem in
     #  https://github.com/sars-cov-2-variants/lineage-proposals/issues/40)
     # The -M 4 allowed up to radius 32, and crazy things started happening all while I was
     # trying to get the tree cleaned up for pango-designation release 1.20 --> lineageTree.
     # After 2023-05-20, when I found that matOptimize had moved a big chunk of B.1 onto a
     # B.1.1.7 garbage branch, causing big trouble for lineageTree (23_05_18_updateLineageTreePb.txt).
     # After that I changed it back to -M 2 for my sanity. If the identical-child thing happens again,
     # then I'll probably just run matOptimize twice, with a small radius the second time.
     cd $ottoDir/$today && $matOptimize \
         -T 64 -m 0.00000001 -M 2 -S move_log.filtered \
         -i merged.deltaMasked.filtered.pb.gz \
         -o gisaidAndPublic.$today.masked.preTrim.pb.gz \
         >& matOptimize.filtered.log
 
     # Fix grandparent-reversion nodes that cause some lineages to be incorrectly placed as
     # sublineages of siblings.
     $matUtils fix -i gisaidAndPublic.$today.masked.preTrim.pb.gz -c 10 \
         -o gisaidAndPublic.$today.masked.preTrim.fix.pb.gz
 
     # Again prune samples with too many private mutations and internal branches that are too long.
     $matUtils extract -i gisaidAndPublic.$today.masked.preTrim.fix.pb.gz \
         --max-parsimony 20 \
-        --max-branch-length 60 \
+        --max-branch-length 70 \
         --max-path-length 225 \
         -O -o gisaidAndPublic.$today.masked.pb.gz
 fi
 
 # Exclude sequences with a very high number of EPPs from future runs
 grep ^Current usher.addNew.log \
 | awk '$16 >= 10 {print $8;}' \
 | awk -F\| '{ if ($3 == "") { print $1; } else { print $2; } }' \
     > tooManyEpps.ids
 cat tooManyEpps.ids >> ../tooManyEpps.ids
 
 $matUtils extract -i gisaidAndPublic.$today.masked.pb.gz -u samples.$today
 
 $scriptDir/combineMetadata.sh $prevDate $today