# for emacs: -*- mode: sh; -*-

# This file describes browser build for the gorGor5
# Organism name:  Gorilla gorilla gorilla (Gorilla)

#  Must find the photograph first, can not continue until finding
#  the photograph.

#########################################################################
#  Initial steps, find photograph (DONE - 2016-06-15 - Hiram)

#  since this is not a new species, can reuse the existing Gorilla photo

# To start this initialBuild.txt document, from a previous assembly document:

mkdir ~/kent/src/hg/makeDb/doc/gorGor5
cd ~/kent/src/hg/makeDb/doc/gorGor5

sed -e 's/colAng1/gorGor5/g; s/ColAng1/GorGor5/g; s/DONE/TBD/g;' ../colAng1/initialBuild.txt > initialBuild.txt

mkdir -p /hive/data/genomes/gorGor5/genbank
cd /hive/data/genomes/gorGor5/genbank

rsync -L -a -P \
rsync://ftp.ncbi.nlm.nih.gov/genomes/genbank/vertebrate_mammalian/Gorilla_gorilla/all_assembly_versions/GCA_900006655.1_GSMRT3/ ./

#  since this is not a new species, can reuse the existing Gorilla photo
# construct the required photoReference.txt
cd /hive/data/genomes/gorGor5
printf "https://commons.wikimedia.org/wiki/User:Arpingstone
Wikimedia Commons/Adrian Pingstone\n" > photoReference.txt

# this information is from the top of
#    genbank/*_assembly_report.txt
#    (aka: genbank/GCA_900006655.1_GSMRT3_assembly_report.txt)

# Assembly name:  GSMRT3
# Organism name:  Gorilla gorilla gorilla (western lowland gorilla)
# Sex:  female
# Taxid:          9595
# BioSample:      SAMEA3541598
# BioProject:     PRJEB10880
# Submitter:      UNIVERSITY OF WASHINGTON
# Date:           2016-3-4
# Assembly type:  haploid
# Release type:   major
# Assembly level: Contig
# Genome representation: full
# WGS project:    CYUI01
# GenBank assembly accession: GCA_900006655.1
#
## Assembly-Units:
## GenBank Unit Accession       RefSeq Unit Accession   Assembly-Unit name
## GCA_900006654.1              Primary Assembly
#
# This assembly only contains unplaced WGS contigs.
# The WGS contigs can be viewed in the WGS Browser:
# https://www.ncbi.nlm.nih.gov/Traces/wgs/?val=CYUI01

#############################################################################
# establish config.ra file (DONE - Hiram - 2016-06-21)
    cd /hive/data/genomes/gorGor5
    ~/kent/src/hg/utils/automation/prepConfig.pl gorGor5 mammal \
      gorilla ./genbank/*_assembly_report.txt > gorGor5.config.ra
# going to need a mitoAcc ?
    # edit result to add note about mito sequence, reuse LN611626.1
    # from gorGor4 and the genBankAccessionID wasn't picked up correctly
    # and common name of Gorilla instead of Western lowland gorilla

    # verify it looks sane
    cat gorGor5.config.ra
# config parameters for makeGenomeDb.pl:
db gorGor5
clade mammal
genomeCladePriority 35
scientificName Gorilla gorilla gorilla
commonName GoriLla
assemblyDate Mar. 2016
assemblyLabel UNIVERSITY OF WASHINGTON
assemblyShortLabel GSMRT3
orderKey 23180
# no mito sequence in this assembly
# gorGor3.1 used refseq sequence: NC_011120.1
# reuse the 'official' one from gorGor4: LN611626.1
mitoAcc LN611626.1
fastaFiles /hive/data/genomes/gorGor5/ucsc/*.fa.gz
# agpFiles /hive/data/genomes/gorGor5/ucsc/*.agp
# there are zero gaps (N's) in these contigs
#  these numbers can be small with no effect:
fakeAgpMinContigGap 1
fakeAgpMinScaffoldGap 100
# qualFiles none
dbDbSpeciesDir gorilla
photoCreditURL https://commons.wikimedia.org/wiki/User:Arpingstone
photoCreditName Wikimedia Commons/Adrian Pingstone
ncbiGenomeId 2156
ncbiAssemblyId 705391
ncbiAssemblyName GSMRT3
ncbiBioProject PRJEB10880
ncbiBioSample SAMEA3541598
genBankAccessionID GCA_900006654.1
taxId 9595

#############################################################################
# setup UCSC named files (DONE - 2016-06-21 - Hiram)

    mkdir /hive/data/genomes/gorGor5/ucsc
    cd /hive/data/genomes/gorGor5/ucsc
    # measure what is in the genbank release:
    faSize ../genbank/GCA_900006655.1_GSMRT3_genomic.fna.gz
# 3080414926 bases (0 N's 3080414926 real 1809316642 upper 1271098284 lower)
#    in 15997 sequences in 1 files
# Total size: mean 192562.0 sd 1464472.7 min 21 (CYUI01004525.1)
#    max 36219563 (CYUI01000001.1) median 12962
# %41.26 masked total, %41.26 masked real

    # check for duplicate sequences:

    time faToTwoBit -noMask ../genbank/*3_genomic.fna.gz genbank.2bit
    #  real    1m14.328s

    twoBitDup genbank.2bit
    # no output is a good result, otherwise, would have to eliminate duplicates

    # there is no AGP for this assembly, just fixup the fasta names,
    # change the .1 to v1
    zcat ../genbank/GCA_900006655.1_GSMRT3_genomic.fna.gz \
       | sed -e 's/.1 Gorilla.*/v1/;' | gzip > ucsc.fa.gz

    # no longer need the 2bit files
    rm genbank.2bit

#############################################################################
#  Initial database build (DONE - 2016-06-21 - Hiram)

    # verify sequence and AGP are OK:
    cd /hive/data/genomes/gorGor5
    time (makeGenomeDb.pl -workhorse=hgwdev -dbHost=hgwdev -fileServer=hgwdev \
         -stop=agp gorGor5.config.ra) > agp.log 2>&1
    # real    2m39.771s

    # then finish it off:
    time (makeGenomeDb.pl -workhorse=hgwdev -dbHost=hgwdev \
       -fileServer=hgwdev -continue=db gorGor5.config.ra) > db.log 2>&1
    # real    31m13.475s

    # check in the trackDb files created and add to trackDb/makefile

    # the fake AGP gets the frag name incorrect for chrM:
    hgsql -e 'update gold set frag="LN611626.1" where chrom="chrM";' gorGor5

    # order key was incorrect:
    hgsql -e 'update dbDb set orderKey=7681 where name="gorGor5";' hgcentraltest

##############################################################################
# cpgIslands on UNMASKED sequence (DONE - 2016-06-21 - Hiram)
    mkdir /hive/data/genomes/gorGor5/bed/cpgIslandsUnmasked
    cd /hive/data/genomes/gorGor5/bed/cpgIslandsUnmasked

    time (doCpgIslands.pl -dbHost=hgwdev -bigClusterHub=ku -buildDir=`pwd` \
       -tableName=cpgIslandExtUnmasked \
          -maskedSeq=/hive/data/genomes/gorGor5/gorGor5.unmasked.2bit \
             -workhorse=hgwdev -smallClusterHub=ku gorGor5) > do.log 2>&1
    # real    7m7.412s

    cat fb.gorGor5.cpgIslandExtUnmasked.txt
    # 31072911 bases of 3080431298 (1.009%) in intersection

#############################################################################
# cytoBandIdeo - (DONE - 2016-06-21 - Hiram)
    mkdir /hive/data/genomes/gorGor5/bed/cytoBand
    cd /hive/data/genomes/gorGor5/bed/cytoBand
    makeCytoBandIdeo.csh gorGor5

#############################################################################
# ucscToINSDC table/track (DONE - 2016-08-05 - Hiram)
    # there are no other names for contigs in this assembly, there are just
    # the contig names

#########################################################################
# fixup search rule for assembly track/gold table (DONE - 2016-08-05 - Hiram)

    cd ~/kent/src/hg/makeDb/trackDb/gorilla/gorGor5
    # preview prefixes and suffixes:
    hgsql -N -e "select frag from gold;" gorGor5 \
      | sed -e 's/[0-9][0-9]*//;' | sort | uniq -c
#    15997 CYUIv1_1
#        1 LN.1

    # implies a search rule of: '[CILNUY0-9]+([v0-9_\.]+)?'

    # verify this rule will find them all or eliminate them all:
    hgsql -N -e "select frag from gold;" gorGor5 | wc -l
    # 15998

    hgsql -N -e "select frag from gold;" gorGor5 \
       | egrep -e '[CILNUY0-9]+([v0-9_\.]+)?' | wc -l
    # 15998

    hgsql -N -e "select frag from gold;" gorGor5 \
       | egrep -v -e '[CILNUY0-9]+([v0-9_\.]+)?' | wc -l
    # 0

    # hence, add to trackDb/rhesus/gorGor5/trackDb.ra
searchTable gold
shortCircuit 1
termRegex [CILNUY0-9]+([v0-9_\.]+)?
query select chrom,chromStart,chromEnd,frag from %s where frag like '%s%%'
searchPriority 8

    # verify searches work in the position box

##########################################################################
# running repeat masker (DONE - 2016-06-21 - Hiram)
    mkdir /hive/data/genomes/gorGor5/bed/repeatMasker
    cd /hive/data/genomes/gorGor5/bed/repeatMasker
    time  (doRepeatMasker.pl -buildDir=`pwd` \
        -bigClusterHub=ku -dbHost=hgwdev -workhorse=hgwdev \
        -smallClusterHub=ku gorGor5) > do.log 2>&1
    # real    1581m32.299s

    cat faSize.rmsk.txt
# 3080431298 bases (0 N's 3080431298 real 1541684463 upper 1538746835 lower)
#    in 15998 sequences in 1 files
# Total size: mean 192551.0 sd 1464427.6 min 21 (CYUI01004525v1)
#    max 36219563 (CYUI01000001v1) median 12964
# %49.95 masked total, %49.95 masked real

    egrep -i "versi|relea" do.log
    # RepeatMasker version open-4.0.5
    #    January 31 2015 (open-4-0-5) version of RepeatMasker
    # CC   RELEASE 20140131;

    time featureBits -countGaps gorGor5 rmsk
    # 1538745117 bases of 3080431298 (49.952%) in intersection
    # real    0m45.618s

    # why is it different than the faSize above ?
    # because rmsk masks out some N's as well as bases, the faSize count above
    #   separates out the N's from the bases, it doesn't show lower case N's

    # with high contig count assemblies, faster way to get the same result:
    time hgsql -N -e 'select genoName,genoStart,genoEnd from rmsk;' gorGor5 \
        | bedSingleCover.pl stdin | ave -col=4 stdin | grep "^total"
    # total 1538745117.000000
    # real    0m39.613s

##########################################################################
# running simple repeat (DONE - 2016-06-21 - Hiram)
    # this was done twice, once with the existing trf v407.b
    # and second with the v4.09
    # there were nearly identical results, a few more bases covered
    # by the v4.09.  Ended up loading the v4.09 results

    mkdir /hive/data/genomes/gorGor5/bed/simpleRepeat
    cd /hive/data/genomes/gorGor5/bed/simpleRepeat
    time (doSimpleRepeat.pl -buildDir=`pwd` -bigClusterHub=ku \
        -trf409=6 -dbHost=hgwdev -workhorse=hgwdev -smallClusterHub=ku \
        gorGor5) > do.log 2>&1
    # real    2409m7.465s

    cat fb.simpleRepeat
    # 313503340 bases of 3080431298 (10.177%) in intersection

    # add to rmsk after it is done:
    cd /hive/data/genomes/gorGor5
    twoBitMask gorGor5.rmsk.2bit \
        -add bed/simpleRepeat/trfMask.bed gorGor5.2bit
    #   you can safely ignore the warning about fields >= 13
    twoBitToFa gorGor5.2bit stdout | faSize stdin > faSize.gorGor5.2bit.txt
    cat faSize.gorGor5.2bit.txt
# 3080431298 bases (0 N's 3080431298 real 1539728819 upper 1540702479 lower)
#    in 15998 sequences in 1 files
# Total size: mean 192551.0 sd 1464427.6 min 21 (CYUI01004525v1)
#    max 36219563 (CYUI01000001v1) median 12964
# %50.02 masked total, %50.02 masked real

    rm /gbdb/gorGor5/gorGor5.2bit
    ln -s `pwd`/gorGor5.2bit /gbdb/gorGor5/gorGor5.2bit

#########################################################################
# CREATE MICROSAT TRACK (DONE - 2016-06-24 - Hiram)
    ssh hgwdev
    mkdir /cluster/data/gorGor5/bed/microsat
    cd /cluster/data/gorGor5/bed/microsat

    awk '($5==2 || $5==3) && $6 >= 15 && $8 == 100 && $9 == 0 {printf("%s\t%s\t%s\t%dx%s\n", $1, $2, $3, $6, $16);}' \
       ../simpleRepeat/simpleRepeat.bed > microsat.bed

    hgLoadBed gorGor5 microsat microsat.bed
    # Read 24885 elements of size 4 from microsat.bed

##########################################################################
## WINDOWMASKER (DONE - 2016-06-23 - Hiram)

    mkdir /hive/data/genomes/gorGor5/bed/windowMasker
    cd /hive/data/genomes/gorGor5/bed/windowMasker
    time (doWindowMasker.pl -buildDir=`pwd` -workhorse=hgwdev \
        -dbHost=hgwdev gorGor5) > do.log 2>&1
    #  real    222m34.996s

    # Masking statistics
    cat faSize.gorGor5.cleanWMSdust.txt
# 3080431298 bases (0 N's 3080431298 real 1790074276 upper 1290357022 lower)
#    in 15998 sequences in 1 files
# Total size: mean 192551.0 sd 1464427.6 min 21 (CYUI01004525v1)
#    max 36219563 (CYUI01000001v1) median 12964
# %41.89 masked total, %41.89 masked real

    cat fb.gorGor5.rmsk.windowmaskerSdust.txt
    # 908347681 bases of 3080431298 (29.488%) in intersection

#########################################################################
# run up idKeys files for ncbiRefSeq (DONE - 2016-06-24 - Hiram)
    mkdir /hive/data/genomes/gorGor5/bed/idKeys
    cd /hive/data/genomes/gorGor5/bed/idKeys

    time (doIdKeys.pl -buildDir=`pwd`  gorGor5) > do.log 2>&1
    # real    18m10.094s

    cat gorGor5.keySignature.txt
    #   a218a7c1e57897febd1a2a7bad7bad97

##########################################################################
# ncbiRefSeq - (DONE - 2016-06-24 - Hiram)
    # this isn't a refseq release, there are no refseq genes

##########################################################################
# cpgIslands - (DONE - 2016-06-24 - Hiram)
    mkdir /hive/data/genomes/gorGor5/bed/cpgIslands
    cd /hive/data/genomes/gorGor5/bed/cpgIslands
    time (doCpgIslands.pl -dbHost=hgwdev -bigClusterHub=ku \
      -workhorse=hgwdev -smallClusterHub=ku gorGor5) > do.log 2>&1
    #   real    8m16.226s

    cat fb.gorGor5.cpgIslandExt.txt
    # 21459917 bases of 3080431298 (0.697%) in intersection

##############################################################################
# genscan - (DONE - 2016-06-24 - Hiram)
    mkdir /hive/data/genomes/gorGor5/bed/genscan
    cd /hive/data/genomes/gorGor5/bed/genscan
    time (doGenscan.pl -buildDir=`pwd` -workhorse=hgwdev -dbHost=hgwdev \
      -bigClusterHub=ku gorGor5) > do.log 2>&1
    # real    68m0.620s

    cat fb.gorGor5.genscan.txt
    # 52474606 bases of 3080431298 (1.703%) in intersection

    cat fb.gorGor5.genscanSubopt.txt
    # 56892790 bases of 3080431298 (1.847%) in intersection

########################################################################
# Create kluster run files (DONE - 2016-06-24 - Hiram)

    # numerator is gorGor5 gapless bases "real" as reported by:
    featureBits -noRandom -noHap gorGor5 gap
    # 0 bases of 3080431298 (0.000%) in intersection


    # denominator is hg19 gapless bases as reported by:
    #   featureBits -noRandom -noHap hg19 gap
    #     234344806 bases of 2861349177 (8.190%) in intersection
    # 1024 is threshold used for human -repMatch:
    calc \( 3080431298 / 2861349177 \) \* 1024
    #  ( 3080431298 / 2861349177 ) * 1024 = 1102.403605

    # ==> use -repMatch=1100 according to size scaled up from 1024 for human.
    #   and rounded down to nearest 50
    cd /hive/data/genomes/gorGor5
    blat gorGor5.2bit \
         /dev/null /dev/null -tileSize=11 -makeOoc=jkStuff/gorGor5.11.ooc \
        -repMatch=1100
    #   Wrote 31384 overused 11-mers to jkStuff/gorGor5.11.ooc

    # there are no non-bridged gaps (there no gaps at all)
    hgsql -e 'select count(*) from gap;' gorGor5
# +----------+
# | count(*) |
# +----------+
# |        0 |
# +----------+

    hgsql -e 'select count(*) from gap where bridge="no";' gorGor5
# +----------+
# | count(*) |
# +----------+
# |        0 |
# +----------+

    # if there were non-bridged gaps, make up a nonBridged.lft file

    #   check non-bridged gaps to see what the typical size is:
#    hgsql -N \
#         -e 'select * from gap where bridge="no" order by size;' gorGor5 \
#         | sort -k7,7nr | ave -col=7 stdin
    #   most non-bridged gaps have size = 100
    #   decide on a minimum gap for this break, use either 100 or 5000 will
    #   generate 13387 liftOver rows, but if use 6000, only got 11703 rows.
    #   so use 100 here to get more liftOver row.
#     gapToLift -verbose=2 -minGap=100 gorGor5 jkStuff/nonBridged.lft \
#         -bedFile=jkStuff/nonBridged.bed

########################################################################
# GENBANK AUTO UPDATE (DONE - 2016-07-22 - Hiram)
    ssh hgwdev
    cd $HOME/kent/src/hg/makeDb/genbank
    git pull
    # /cluster/data/genbank/data/organism.lst shows:
    # #organism     mrnaCnt estCnt  refSeqCnt
    # Gorilla gorilla 591     0       88
    # Gorilla gorilla gorilla 8       0       0

    # Edit src/lib/gbGenome.c to add new species

    # edit etc/genbank.conf to add gorGor5 just after balAcu1

# Gorilla
gorGor5.serverGenome = /hive/data/genomes/gorGor5/gorGor5.2bit
gorGor5.clusterGenome = /hive/data/genomes/gorGor5/gorGor5.2bit
gorGor5.ooc = /hive/data/genomes/gorGor5/jkStuff/gorGor5.11.ooc
gorGor5.lift = no
gorGor5.perChromTables = no
gorGor5.refseq.mrna.native.pslCDnaFilter  = ${ordered.refseq.mrna.native.pslCDnaFilter}
gorGor5.refseq.mrna.xeno.pslCDnaFilter    = ${ordered.refseq.mrna.xeno.pslCDnaFilter}
gorGor5.genbank.mrna.native.pslCDnaFilter = ${ordered.genbank.mrna.native.pslCDnaFilter}
gorGor5.genbank.mrna.xeno.pslCDnaFilter   = ${ordered.genbank.mrna.xeno.pslCDnaFilter}
gorGor5.genbank.est.native.pslCDnaFilter  = ${ordered.genbank.est.native.pslCDnaFilter}
gorGor5.genbank.est.xeno.pslCDnaFilter    = ${ordered.genbank.est.xeno.pslCDnaFilter}
gorGor5.downloadDir = gorGor5
# default yes refseq.mrna.native refseq.mrna.xeno genbank.mrna.native
# default yes genbank.est.native
# default no genbank.mrna.xeno genbank.est.xeno

    git commit -m "Added gorGor5 - Gorilla refs #17580" etc/genbank.conf
    git push
    # update /cluster/data/genbank/:
    make etc-update

    screen      #  control this business with a screen since it takes a while
    cd /cluster/data/genbank

    time ./bin/gbAlignStep -initial gorGor5
    # real    67m3.917s

    # logFile: var/build/logs/2016.07.05-12:53:02.gorGor5.initalign.log
    # tail -2 var/build/logs/2016.07.05-12:53:02.gorGor5.initalign.log
# hgwdev 2016.07.05-13:59:57 gorGor5.initalign: Succeeded: gorGor5
# hgwdev 2016.07.05-14:00:06 gorGor5.initalign: finish

    #   To re-do, rm the dir first:
    #     /cluster/data/genbank/work/initial.gorGor5

    # load database when finished
    ssh hgwdev
    cd /cluster/data/genbank
    time ./bin/gbDbLoadStep -drop -initialLoad gorGor5
    # real    18m33.387s
    tail -1 var/dbload/hgwdev/logs/2016.07.21-23:42:20.gorGor5.dbload.log
# hgwdev 2016.07.22-00:00:53 gorGor5.dbload: finish

    # enable daily alignment and update of hgwdev
    cd ~/kent/src/hg/makeDb/genbank
    git pull
    # add gorGor5 to:
    #   etc/align.dbs etc/hgwdev.dbs
    git add etc/align.dbs etc/hgwdev.dbs
    git commit -m "Added gorGor5 - Gorilla  - refs #17580" etc/align.dbs etc/hgwdev.dbs
    git push
    make etc-update

#############################################################################
# augustus gene track (DONE - 2016-06-24 - Hiram)

    mkdir /hive/data/genomes/gorGor5/bed/augustus
    cd /hive/data/genomes/gorGor5/bed/augustus
    time (doAugustus.pl -buildDir=`pwd` -bigClusterHub=ku \
        -species=human -dbHost=hgwdev \
           -workhorse=hgwdev gorGor5) > do.log 2>&1
    # real    200m53.178s

    cat fb.gorGor5.augustusGene.txt
    # 55196723 bases of 3080431298 (1.792%) in intersection

############################################################################
# construct liftOver to gorGor4 (DONE - 2016-06-24,26 - Hiram)
    screen -S gorGor4	# manage this longish running job in a screen
    mkdir /hive/data/genomes/gorGor5/bed/blat.gorGor4.2016-06-24
    cd /hive/data/genomes/gorGor5/bed/blat.gorGor4.2016-06-24
    # check it with -debug first to see if it is going to work:
    time doSameSpeciesLiftOver.pl -buildDir=`pwd` -bigClusterHub=ku \
	-ooc=/hive/data/genomes/gorGor5/jkStuff/gorGor5.11.ooc \
	-debug -dbHost=hgwdev -workhorse=hgwdev gorGor5 gorGor4
    # real    0m2.154s

    # if that is OK, then run it:
    time (doSameSpeciesLiftOver.pl -buildDir=`pwd` -bigClusterHub=ku \
	-ooc=/hive/data/genomes/gorGor5/jkStuff/gorGor5.11.ooc \
	-dbHost=hgwdev -workhorse=hgwdev gorGor5 gorGor4) > do.log 2>&1
    #	real    2412m36.913s

    # verify this file exists:
    #	/gbdb/gorGor5/liftOver/gorGor5ToGorGor4.over.chain.gz
    # and try out the conversion on genome-test from gorGor5 to gorGor4

############################################################################
# construct liftOver gorGor4 to gorGor5 (DONE - 2016-06-24,26 - Hiram)
    screen -S gorGor4	# manage this longish running job in a screen
    mkdir /hive/data/genomes/gorGor4/bed/blat.gorGor5.2016-06-24
    cd /hive/data/genomes/gorGor4/bed/blat.gorGor5.2016-06-24
    # check it with -debug first to see if it is going to work:
    time doSameSpeciesLiftOver.pl -buildDir=`pwd` -bigClusterHub=ku \
	-ooc=/hive/data/genomes/gorGor4/jkStuff/gorGor4.11.ooc \
	-debug -dbHost=hgwdev -workhorse=hgwdev gorGor4 gorGor5
    # real    0m1.954s

    # if that is OK, then run it:
    time (doSameSpeciesLiftOver.pl -buildDir=`pwd` -bigClusterHub=ku \
	-ooc=/hive/data/genomes/gorGor4/jkStuff/gorGor4.11.ooc \
	-dbHost=hgwdev -workhorse=hgwdev gorGor4 gorGor5) > do.log 2>&1
    #	real    3222m59.287s

    # verify this file exists:
    #	/gbdb/gorGor4/liftOver/gorGor4ToGorGor5.over.chain.gz
    # and try out the conversion on genome-test from gorGor4 to gorGor5

############################################################################
#  BLATSERVERS ENTRY (DONE - 2016-08-05 - Hiram)
#	After getting a blat server assigned by the Blat Server Gods,
    ssh hgwdev

    hgsql -e 'INSERT INTO blatServers (db, host, port, isTrans, canPcr) \
	VALUES ("gorGor5", "blat1d", "17868", "1", "0"); \
	INSERT INTO blatServers (db, host, port, isTrans, canPcr) \
	VALUES ("gorGor5", "blat1d", "17869", "0", "1");' \
	    hgcentraltest
    #	test it with some sequence

############################################################################
## default position same as gorGor4 EVPL gene
#                                 (DONE - 2016-08-04 - Hiram)
    ssh hgwdev
    hgsql -e 'update dbDb set defaultPos="CYUI01015056v1:3670206-3687174"
	where name="gorGor5";' hgcentraltest

#########################################################################
# all.joiner update, downloads and in pushQ - (DONE - 2016-08-08 - Hiram)
    cd $HOME/kent/src/hg/makeDb/schema
    # fixup all.joiner until this is a clean output
    joinerCheck -database=gorGor5 -tableCoverage all.joiner
    joinerCheck -database=gorGor5 -times all.joiner
    joinerCheck -database=gorGor5 -keys all.joiner

    cd /hive/data/genomes/gorGor5
    time (makeDownloads.pl gorGor5) > downloads.log 2>&1
    #  real    19m52.657s

    #   now ready for pushQ entry
    mkdir /hive/data/genomes/gorGor5/pushQ
    cd /hive/data/genomes/gorGor5/pushQ
    time makePushQSql.pl gorGor5 > gorGor5.pushQ.sql 2> stderr.out
    #   real    11m56.179s

    #   check for errors in stderr.out, some are OK, e.g.:
    # WARNING: gorGor5 does not have ucscToINSDC
    # WARNING: hgwdev does not have /gbdb/gorGor5/wib/gc5Base.wib
    # WARNING: hgwdev does not have /gbdb/gorGor5/wib/quality.wib
    # WARNING: hgwdev does not have /gbdb/gorGor5/bbi/qualityBw/quality.bw
    # WARNING: gorGor5 does not have seq
    # WARNING: gorGor5 does not have extFile
    # WARNING: gorGor5 does not have estOrientInfo

    #   copy it to hgwbeta
    scp -p gorGor5.pushQ.sql qateam@hgwbeta:/tmp/
    ssh qateam@hgwbeta "./bin/x86_64/hgsql qapushq < /tmp/gorGor5.pushQ.sql"

    #   in that pushQ entry walk through each entry and see if the
    #   sizes will set properly

#########################################################################
# LIFTOVER TO gorGor6 (DONE - 2019-11-20 - Hiram)
    ssh hgwdev
    mkdir /hive/data/genomes/gorGor5/bed/blat.gorGor6.2019-11-20
    cd /hive/data/genomes/gorGor5/bed/blat.gorGor6.2019-11-20
    doSameSpeciesLiftOver.pl -verbose=2 \
        -debug -bigClusterHub=ku -dbHost=hgwdev -workhorse=hgwdev \
        -ooc=/hive/data/genomes/gorGor5/jkStuff/gorGor5.11.ooc \
         gorGor5 gorGor6
    time (doSameSpeciesLiftOver.pl -verbose=2 \
        -bigClusterHub=ku -dbHost=hgwdev -workhorse=hgwdev \
        -ooc=/hive/data/genomes/gorGor5/jkStuff/gorGor5.11.ooc \
         gorGor5 gorGor6) > doLiftOverToGorGor6.log 2>&1
    # real    697m29.645s

    # see if the liftOver menus function in the browser from gorGor5 to gorGor6

#########################################################################
# ucscToINSDC and ucscToRefSeq table/track (DONE - 2021-07-02 - Hiram)
    # construct idKeys for the genbank sequence

    mkdir -p /hive/data/genomes/gorGor5/genbank/idKeys
    cd /hive/data/genomes/gorGor5/genbank/idKeys
    faToTwoBit ../GCA_900006655.1_GSMRT3_genomic.fna.gz gorGor5.genbank.2bit

    time (doIdKeys.pl -buildDir=`pwd` \
        -twoBit=`pwd`/gorGor5.genbank.2bit gorGor5Genbank) > do.log 2>&1 &
    # real    0m41.551s


    cat gorGor5Genbank.keySignature.txt
    #  9939b7dbae923d899d2fe82dccf7ebd0

    mkdir /hive/data/genomes/gorGor5/bed/chromAlias
    cd /hive/data/genomes/gorGor5/bed/chromAlias

    join -t$'\t' ../idKeys/gorGor5.idKeys.txt \
        ../../genbank/idKeys/gorGor5Genbank.idKeys.txt | cut -f2- \
          | sort -k1,1 | join -t$'\t' <(sort -k1,1 ../../chrom.sizes) - \
            | awk '{printf "%s\t0\t%d\t%s\n", $1, $2, $3}' \
               | sort -k1,1 -k2,2n > ucscToINSDC.bed

    # and the chrM genbank name and size is:
    grep chrM ../../*.agp | cut -f3,6
    #	16372   LN611626.1

    printf "chrM\t0\t16372\tLN611626.1\n" >> ucscToINSDC.bed

    # should be same line counts throughout:
    wc -l * ../../chrom.sizes
    #	15998 ucscToINSDC.bed
    #	15998 ../../chrom.sizes

    export chrSize=`cut -f1 ucscToINSDC.bed | awk '{print length($0)}' | sort -n | tail -1`
    echo $chrSize
    # 14
    # use the $chrSize in this sed
    sed -e "s/21/$chrSize/" $HOME/kent/src/hg/lib/ucscToINSDC.sql \
         | hgLoadSqlTab gorGor5 ucscToINSDC stdin ucscToINSDC.bed

    # should be quiet for all OK
    checkTableCoords gorGor5 ucscToINSDC

    # should cover %100 entirely:
    featureBits -countGaps gorGor5 ucscToINSDC
    #  3080431298 bases of 3080431298 (100.000%) in intersection

#########################################################################
# add chromAlias table (DONE - 2021-07-02 - Hiram)

    mkdir /hive/data/genomes/gorGor5/bed/chromAlias
    cd /hive/data/genomes/gorGor5/bed/chromAlias


    hgsql -N -e 'select chrom,name from ucscToINSDC;' gorGor5 \
        | sort -k1,1 > ucsc.genbank.tab

    wc -l * ../../chrom.sizes
    #	15998 ucsc.genbank.tab
    #	15998 ucscToINSDC.bed
    #	15998 ../../chrom.sizes

    # make up an MT entry, call it 'assembly' type:
    printf "chrM\tMT\n" > ucsc.assembly.tab

    ~/kent/src/hg/utils/automation/chromAlias.pl ucsc.*.tab \
        > gorGor5.chromAlias.tab

for t in genbank assembly
do
  c0=`cat ucsc.$t.tab | wc -l`
  c1=`grep $t gorGor5.chromAlias.tab | wc -l`
  ok="OK"
  if [ "$c0" -ne "$c1" ]; then
     ok="ERROR"
  fi
  printf "# checking $t: $c0 =? $c1 $ok\n"
done
# checking genbank: 15998 =? 15998 OK
# checking assembly: 1 =? 1 OK

    # verify chrM is here properly:
    grep chrM gorGor5.chromAlias.tab
# LN611626.1      chrM    genbank
# MT      chrM    assembly

    hgLoadSqlTab gorGor5 chromAlias ~/kent/src/hg/lib/chromAlias.sql \
        gorGor5.chromAlias.tab

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