##############################################################################
# LASTZ sheep GCA_002844635.1 (DONE - 2021-04-20 - Hiram)
    mkdir /hive/data/genomes/susScr11/bed/lastzGCA_002844635.1.2021-04-20
    cd /hive/data/genomes/susScr11/bed/lastzGCA_002844635.1.2021-04-20

    printf '# GCA_002844635.1 vs hg38
BLASTZ=/cluster/bin/penn/lastz-distrib-1.04.03/bin/lastz
BLASTZ_T=2
BLASTZ_O=400
BLASTZ_E=30
BLASTZ_M=254
# default BLASTZ_Q score matrix:
#       A     C     G     T
# A    91  -114   -31  -123
# C  -114   100  -125   -31
# G   -31  -125   100  -114
# T  -123   -31  -114    91

# TARGET: Human Hg38
SEQ1_DIR=/hive/data/genomes/hg38/hg38.2bit
SEQ1_LEN=/hive/data/genomes/hg38/chrom.sizes
SEQ1_CHUNK=20000000
SEQ1_LAP=10000

# QUERY: sheep GCA_002844635.1
SEQ2_DIR=/hive/data/genomes/asmHubs/GCF/002/742/125/GCA_002844635.1/GCA_002844635.1.2bit
SEQ2_LEN=/hive/data/genomes/asmHubs/GCF/002/742/125/GCA_002844635.1/GCA_002844635.1.chrom.sizes.txt
SEQ2_CHUNK=20000000
SEQ2_LAP=0
SEQ2_LIMIT=100

BASE=/hive/data/genomes/hg38/bed/lastzGCA_002844635.1.2021-04-20
TMPDIR=/dev/shm
' > DEF

export targetDb="hg38"
export asmId="GCA_002844635.1"
export gcPath="GCA/002/844/635"
cd /hive/data/genomes/$targetDb/bed/lastz${asmId}.2021-04-20
time (doBlastzChainNet.pl -trackHub -noDbNameCheck -verbose=2 `pwd`/DEF \
   -workhorse=hgwdev -smallClusterHub=hgwdev -bigClusterHub=ku \
     -syntenicNet -chainMinScore=3000 -chainLinearGap=medium) > do.log 2>&1
cat fb.${targetDb}.chain.${asmId}Link.txt
cat fb.${targetDb}.chainSyn.${asmId}Link.txt

grep -w real do.log | sed -e 's/^/    # /;'
    # real      318m24.582s

sed -e 's/^/    # /;' fb.$targetDb.chain.${asmId}Link.txt
    # 2431952924 bases of 2501912388 (97.204%) in intersection
sed -e 's/^/    # /;' fb.$targetDb.chainSyn.${asmId}Link.txt
    # 2409981908 bases of 2501912388 (96.326%) in intersection

time (doRecipBest.pl -trackHub -load -workhorse=hgwdev -buildDir=`pwd` \
-query2Bit="/hive/data/genomes/asmHubs/$gcPath/${asmId}/${asmId}.2bit" \
-querySizes="/hive/data/genomes/asmHubs/$gcPath/${asmId}/${asmId}.chrom.sizes.txt" \
$targetDb ${asmId}) >> rbest.log 2>&1
grep -w real rbest.log | sed -e 's/^/    # /;'
    # real      93m2.731s

sed -e 's/^/    # /;' fb.$targetDb.chainRBest.$asmId.txt
    # 2381230203 bases of 2501912388 (95.176%) in intersection

## Overall time for the above:
# real    411m27.346s

### the swap to the assembly hub
export target="susScr11"
export Target="SusScr11"
export query="GCA_002844635.1"
export asmId="GCA_002844635.1_USMARCv1.0"
export gcPath="GCA/002/844/635"

mkdir -p /hive/data/genomes/asmHubs/refseqBuild/$gcPath/$asmId/trackData/blastz.$target.swap
cd /hive/data/genomes/asmHubs/refseqBuild/$gcPath/$asmId/trackData/blastz.$target.swap

time (doBlastzChainNet.pl -trackHub -noDbNameCheck -verbose=2 -swapDir=`pwd` \
    /hive/data/genomes/${target}/bed/lastz.${query}/DEF -syntenicNet \
  -workhorse=hgwdev -smallClusterHub=hgwdev -bigClusterHub=ku \
    -swap -chainMinScore=3000 -chainLinearGap=medium) >> swap.log 2>&1
grep -w real swap.log | sed -e 's/^/    # /;'
    # real      90m38.556s

sed -e 's/^/    # /;' fb.${query}.chain.${Target}Link.txt
    # 2543057365 bases of 2755438182 (92.292%) in intersection
sed -e 's/^/    # /;' fb.${query}.chainSyn.${Target}Link.txt
    # 2514166037 bases of 2755438182 (91.244%) in intersection

time (doRecipBest.pl -trackHub -load -workhorse=hgwdev -buildDir=`pwd` \
-target2Bit="/hive/data/genomes/asmHubs/$gcPath/${query}/${query}.2bit" \
-targetSizes="/hive/data/genomes/asmHubs/$gcPath/${query}/${query}.chrom.sizes.txt" \
$query $target) >> rbest.log 2>&1
grep -w real rbest.log | sed -e 's/^/    # /;'
    # real      86m43.960s

sed -e 's/^/    # /;' fb.${query}.chainRBest.${Target}.txt
    # 2388751543 bases of 2755438182 (86.692%) in intersection

# Complete run time for all the swap operation:
    # real    285m1.340s

##############################################################################