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

# Drosophila erecta -- Agencourt "CAF1" via Eisen's 12-fly site

# THIS IS ONLY TO GET MASKED SEQUENCE -- NOT A BROWSER AT THIS POINT


#########################################################################
# DOWNLOAD SEQUENCE (DONE 9/20/06 angie)
    ssh kkstore05
    mkdir /cluster/store12/droEre2
    ln -s /cluster/store12/droEre2 /cluster/data/droEre2
    mkdir /cluster/data/droEre2/downloads
    cd /cluster/data/droEre2/downloads
    wget http://rana.lbl.gov/drosophila/caf1/dere_caf1.tar.gz
    tar xvzf dere_caf1.tar.gz
    cd dere
    faSize scaffolds.bases
#152712140 bases (7628121 N's 145084019 real 145084019 upper 0 lower) in 5124 sequences in 1 files
#Total size: mean 29803.3 sd 721223.0 min 154 (scaffold_5056) max 26641161 (scaffold_4929) median 1729
#N count: mean 1488.7 sd 9323.5
#U count: mean 28314.6 sd 717773.6
#L count: mean 0.0 sd 0.0


#########################################################################
# MAKE GENOME DB *UP TO DB STEP ONLY* (DONE 9/20/06 angie)
    ssh kkstore05
    cd /cluster/data/droEre2
    cat > droEre2.config.ra <<EOF
# Config parameters for makeGenomeDb.pl:
db droEre2
clade insect
scientificName Drosophila erecta
assemblyDate Feb. 2006
assemblyLabel Agencourt CAF1
orderKey 53
mitoAcc none
fastaFiles /cluster/data/droEre2/downloads/dere/scaffolds.bases
agpFiles /cluster/data/droEre2/downloads/dere/assembly.agp
dbDbSpeciesDir drosophila
EOF

    # Stop at db step so we can use featureBits, but don't do dbDb and trackDb
    # because we're not building an actual browser for now.
    makeGenomeDb.pl droEre2.config.ra -stop=db \
      >& makeGenomeDb.log & tail -f makeGenomeDb.log


#########################################################################
# REPEATMASKER (DONE 9/20/06 angie)
    ssh kkstore05
    # Run -debug to create the dir structure and preview the scripts:
    doRepeatMasker.pl droEre2 -verbose 3 -debug
    # Run it for real and tail the log:
    doRepeatMasker.pl droEre2 -species drosophila -verbose 3 \
      >& /cluster/data/droEre2/bed/RepeatMasker.2006-09-20/do.log &
    tail -f /cluster/data/droEre2/bed/RepeatMasker.2006-09-20/do.log
    # RepeatMasker and lib version from do.log:
#    March 20 2006 (open-3-1-5) version of RepeatMasker
#CC   RELEASE 20060315;                                            *
    # Compare coverage to previous assembly:
    featureBits droEre2 rmsk
#18539872 bases of 145084019 (12.779%) in intersection
    featureBits droEre1 rmsk
#18339464 bases of 145196048 (12.631%) in intersection


#########################################################################
# SIMPLE REPEATS (TRF) (DONE 9/20/06 angie)
    ssh kolossus
    nice tcsh
    mkdir /cluster/data/droEre2/bed/simpleRepeat
    cd /cluster/data/droEre2/bed/simpleRepeat
    twoBitToFa ../../droEre2.unmasked.2bit stdout \
    | trfBig -trf=/cluster/bin/i386/trf stdin /dev/null \
      -bedAt=simpleRepeat.bed -tempDir=/tmp \
    >& trf.log & tail -f trf.log
    # ~52 minutes (longer than D. mel, must be because of the scaffolds)

    # Make a filtered version for sequence masking:
    awk '{if ($5 <= 12) print;}' simpleRepeat.bed > trfMask.bed

    # Load unfiltered repeats into the database:
    ssh hgwdev
    hgLoadBed droEre2 simpleRepeat \
      /cluster/data/droEre2/bed/simpleRepeat/simpleRepeat.bed \
      -sqlTable=$HOME/kent/src/hg/lib/simpleRepeat.sql
    # Compare coverage to previous assembly:
    featureBits droEre2 simpleRepeat
#9028488 bases of 145084019 (6.223%) in intersection
    featureBits droEre1 simpleRepeat
#9071572 bases of 145196048 (6.248%) in intersection


#########################################################################
# MASK SEQUENCE WITH FILTERED TRF IN ADDITION TO RM (DONE 9/20/06 angie)
    ssh kolossus
    cd /cluster/data/droEre2
    time twoBitMask droEre2.rmsk.2bit -add bed/simpleRepeat/trfMask.bed droEre2.2bit
    # This warning is OK -- the extra fields are not block coordinates.
#Warning: BED file bed/simpleRepeat/trfMask.bed has >=13 fields which means it might contain block coordinates, but this program uses only the first three fields (the entire span -- no support for blocks).
#0.161u 0.368s 0:01.61 32.2%     0+0k 0+0io 1pf+0w

    # Because this is a no-browser build (just masking for alignment)
    # I did not make the usual /gbdb/$db/$db.2bit link.


#########################################################################
# SWAP DM3 CHAIN/NET (DONE 4/3/09 angie)
    mkdir /hive/data/genomes/droEre2/bed/blastz.dm3.swap
    cd /hive/data/genomes/droEre2/bed/blastz.dm3.swap
    doBlastzChainNet.pl -swap -bigClusterHub swarm -smallClusterHub memk \
      -workhorse kolossus \
      /hive/data/genomes/dm3/bed/blastz.droEre2/DEF >& do.log &
    tail -f do.log
    ln -s blastz.dm3.swap /hive/data/genomes/droEre2/bed/blastz.dm3


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