# for emacs: -*- mode: sh; -*- # Drosophila sechellia -- # # Broad's 28 Oct. 2005 assembly # # NOTE: this doc may have genePred loads that fail to include # the bin column. Please correct that for the next build by adding # a bin column when you make any of these tables: # # mysql> SELECT tableName, type FROM trackDb WHERE type LIKE "%Pred%"; # +-------------+---------------------------------+ # | tableName | type | # +-------------+---------------------------------+ # | xenoRefGene | genePred xenoRefPep xenoRefMrna | # | genscan | genePred genscanPep | # +-------------+---------------------------------+ # DOWNLOAD SEQUENCE (DONE 11/29/05 angie) ssh kkstore02 mkdir /cluster/store11/droSec1 cd /cluster/data ln -s /cluster/store11/droSec1 droSec1 cd /cluster/data/droSec1 mkdir jkStuff bed mkdir downloads cd downloads wget http://www.broad.mit.edu/ftp/pub/assemblies/insects/droSec1/assembly.agp wget http://www.broad.mit.edu/ftp/pub/assemblies/insects/droSec1/assembly.links wget http://www.broad.mit.edu/ftp/pub/assemblies/insects/droSec1/BasicAssemblyOneLiner.out wget http://www.broad.mit.edu/ftp/pub/assemblies/insects/droSec1/assembly.quals.gz wget http://www.broad.mit.edu/ftp/pub/assemblies/insects/droSec1/scaffolds.fasta mv scaffolds.fasta ../scaffolds.fa faSize ../scaffolds.fa #166577145 bases (9338570 N's 157238575 real 157238575 upper 0 lower) in 14730 sequences in 1 files #Total size: mean 11308.7 sd 253114.3 min 204 (super_14729) max 21120651 (super_0) median 1710 #N count: mean 634.0 sd 6174.5 #U count: mean 10674.7 sd 248396.0 #L count: mean 0.0 sd 0.0 # PARTITION SCAFFOLDS FOR REPEATMASKER RUN (DONE 11/29/05 angie) # Chop up large scaffolds into ~500kb chunks for RepeatMasker, # then glom the tiny scaffolds up into ~500k collections (looks like # some almost-500k pieces are glommed together --> a few almost-1M chunks, # but that's OK). ssh kkstore02 cd /cluster/data/droSec1 mkdir scaffoldsSplit faSplit size scaffolds.fa 500000 -oneFile \ scaffoldsSplit -lift=jkStuff/scaffoldsSplit.lft mkdir chunks500k faSplit about scaffoldsSplit.fa 500000 chunks500k/chunk_ # CREATING DATABASE (DONE 11/29/05 angie) # Create the database. ssh hgwdev # Make sure there is at least 5 gig free for the database df -h /var/lib/mysql #/dev/sdc1 1.8T 1.3T 356G 79% /var/lib/mysql hgsql '' -e 'create database droSec1' # Copy the table "grp" from an existing database to the new database hgsql droSec1 -e 'create table grp (PRIMARY KEY(NAME)) select * from dm2.grp' # RUN REPEAT MASKER (DONE 11/30/05 angie) ssh kkstore02 cat /cluster/bluearc/RepeatMasker/Libraries/version #RepBase Update 9.11, RM database version 20050112 # make the run directory, output directory, and job list cd /cluster/data/droSec1 cat << '_EOF_' > jkStuff/RMDrosophila #!/bin/csh -fe cd $1 /bin/mkdir -p /tmp/droSec1/$2 /bin/cp ../chunks500k/$2 /tmp/droSec1/$2/ pushd /tmp/droSec1/$2 /cluster/bluearc/RepeatMasker/RepeatMasker -s -spec drosophila $2 popd /bin/cp /tmp/droSec1/$2/$2.out ./ /bin/rm -fr /tmp/droSec1/$2/* /bin/rmdir --ignore-fail-on-non-empty /tmp/droSec1/$2 /bin/rmdir --ignore-fail-on-non-empty /tmp/droSec1 '_EOF_' # << this line makes emacs coloring happy chmod +x jkStuff/RMDrosophila mkdir RMRun RMOut cp /dev/null RMRun/RMJobs foreach f ( chunks500k/*.fa ) set chunk = $f:t echo ../jkStuff/RMDrosophila \ /cluster/data/droSec1/RMOut $chunk \ '{'check in line+ /cluster/data/droSec1/$f'}' \ '{'check out line+ /cluster/data/droSec1/RMOut/$chunk.out'}' \ >> RMRun/RMJobs end # do the run ssh kk cd /cluster/data/droSec1/RMRun para make RMJobs para time #Completed: 312 of 312 jobs #Average job time: 7359s 122.64m 2.04h 0.09d #Longest finished job: 15164s 252.73m 4.21h 0.18d #Submission to last job: 15165s 252.75m 4.21h 0.18d # Lift up the split-scaffold .out's to scaffold .out's ssh kkstore02 cd /cluster/data/droSec1 foreach f (RMOut/*.fa.out) liftUp $f:r:r.scaf.out jkStuff/scaffoldsSplit.lft warn $f > /dev/null end # Make a consolidated scaffold .out file too: head -3 RMOut/chunk_00.fa.out > RMOut/scaffolds.fa.out foreach f (RMOut/*.scaf.out) tail +4 $f >> RMOut/scaffolds.fa.out end # Load the .out files into the database with: ssh hgwdev hgLoadOut -nosplit droSec1 /cluster/data/droSec1/RMOut/scaffolds.fa.out #note: 3 records dropped due to repStart > repEnd #bad rep range [868, 862] line 42582 of /cluster/data/droSec1/RMOut/scaffolds.fa.out #bad rep range [865, 862] line 49234 of /cluster/data/droSec1/RMOut/scaffolds.fa.out #bad rep range [862, 861] line 49837 of /cluster/data/droSec1/RMOut/scaffolds.fa.out # Sent test case to Robert Hubley. # GOLD, GAP FROM AGP (DONE 11/30/05 angie) ssh hgwdev cd /cluster/data/droSec1 # Weird, their sequences are called "super_*" but their agp has # "scaffold_*"... go with the sequence names and fix the agp. sed -e 's/^scaffold_/super_/' downloads/assembly.agp > ucsc.agp hgGoldGapGl -noGl droSec1 ucsc.agp # SIMPLE REPEATS (TRF) (DONE 11/30/05 angie) ssh kolossus mkdir /cluster/data/droSec1/bed/simpleRepeat cd /cluster/data/droSec1/bed/simpleRepeat nice trfBig -trf=/cluster/bin/i386/trf ../../scaffolds.fa \ /dev/null -bedAt=simpleRepeat.bed -tempDir=/tmp \ |& egrep -v '^(Removed|Tandem|Copyright|Loading|Allocating|Initializing|Computing|Scanning|Freeing)' \ > trf.log & # check on this with tail -f trf.log # Load this into the database as so ssh hgwdev hgLoadBed droSec1 simpleRepeat \ /cluster/data/droSec1/bed/simpleRepeat/simpleRepeat.bed \ -sqlTable=$HOME/kent/src/hg/lib/simpleRepeat.sql # FILTER SIMPLE REPEATS (TRF) INTO MASK (DONE 11/30/05 angie) # make a filtered version of the trf output: # keep trf's with period <= 12: ssh kkstore02 cd /cluster/data/droSec1/bed/simpleRepeat awk '{if ($5 <= 12) print;}' simpleRepeat.bed > trfMask.bed # MASK FA USING REPEATMASKER AND FILTERED TRF FILES (DONE 11/30/05 angie) ssh kkstore02 cd /cluster/data/droSec1 maskOutFa -soft scaffolds.fa bed/simpleRepeat/trfMask.bed \ scaffolds.fa maskOutFa -softAdd scaffolds.fa RMOut/scaffolds.fa.out scaffolds.fa # Now clean up the unmasked split scaffolds to avoid confusion later. rm -r chunks500k scaffoldsSplit.fa jkStuff/scaffoldsSplit.lft # STORE SEQUENCE AND ASSEMBLY INFORMATION (DONE 11/30/05 angie) # Translate to 2bit ssh kkstore02 cd /cluster/data/droSec1 faToTwoBit scaffolds.fa droSec1.2bit # Make chromInfo.tab. mkdir bed/chromInfo twoBitInfo droSec1.2bit stdout \ | awk '{printf "%s\t%s\t/gbdb/droSec1/droSec1.2bit\n", $1, $2;}' \ > bed/chromInfo/chromInfo.tab # Make symbolic a link from /gbdb/droSec1 to the 2bit. ssh hgwdev mkdir -p /gbdb/droSec1 ln -s /cluster/data/droSec1/droSec1.2bit /gbdb/droSec1/ # Load chromInfo table. hgsql droSec1 < $HOME/kent/src/hg/lib/chromInfo.sql hgsql droSec1 -e 'load data local infile \ "/cluster/data/droSec1/bed/chromInfo/chromInfo.tab" into table chromInfo' # Make chrom.sizes from chromInfo contents and check scaffold count. hgsql droSec1 -N -e 'select chrom,size from chromInfo' \ > /cluster/data/droSec1/chrom.sizes wc -l /cluster/data/droSec1/chrom.sizes # 14730 /cluster/data/droSec1/chrom.sizes # MAKE HGCENTRALTEST ENTRY AND TRACKDB TABLE (DONE 11/30/05 angie) # Make trackDb table so browser knows what tracks to expect: ssh hgwdev cd ~/kent/src/hg/makeDb/trackDb cvs up -d -P # Edit trackDb/makefile to add droSec1 to the DBS variable. mkdir drosophila/droSec1 # Create a simple drosophila/droSec1/description.html file. cvs add drosophila/droSec1 cvs add drosophila/droSec1/description.html make update DBS=droSec1 # go public on genome-test cvs ci -m "Added droSec1." makefile cvs ci -m "Elementary description.html for droSec1." drosophila/droSec1 mkdir /gbdb/droSec1/html # in a clean, updated tree's kent/src/hg/makeDb/trackDb: make alpha # Note: genome and organism fields must correspond # with defaultDb values hgsql -h genome-testdb hgcentraltest -e \ 'INSERT INTO dbDb \ (name, description, nibPath, organism, \ defaultPos, active, orderKey, genome, scientificName, \ htmlPath, hgNearOk, hgPbOk, sourceName) values \ ("droSec1", "Oct. 2005", "/gbdb/droSec1", "D. sechellia", \ "super_14:795001-815000", 1, 52, \ "D. sechellia", \ "Drosophila sechellia", "/gbdb/droSec1/html/description.html", \ 0, 0, "Broad 28 Oct. 2005");' echo 'INSERT INTO defaultDb (genome, name) values ("D. sechellia", "droSec1");' \ | hgsql -h genome-testdb hgcentraltest hgsql -h genome-testdb hgcentraltest \ -e 'INSERT INTO genomeClade (genome, clade, priority) values \ ("D. sechellia", "insect", 18);' # MAKE GCPERCENT (DONE 11/30/05 angie) ssh hgwdev mkdir /cluster/data/droSec1/bed/gc5Base cd /cluster/data/droSec1/bed/gc5Base hgGcPercent -wigOut -doGaps -file=stdout -win=5 -verbose=2 droSec1 \ /cluster/data/droSec1 | wigEncode stdin gc5Base.wig gc5Base.wib mkdir /gbdb/droSec1/wib ln -s `pwd`/gc5Base.wib /gbdb/droSec1/wib hgLoadWiggle -pathPrefix=/gbdb/droSec1/wib droSec1 gc5Base gc5Base.wig # PUT SEQUENCE ON /ISCRATCH FOR BLASTZ AND MAKE 11.OOC (DONE 11/30/05 angie) # First, agglomerate small scaffolds into chunks of ~100k median # (many scaffolds are larger than that) so we don't have too many # files for one dir, but keep a reasonably low job run time: ssh kkstore02 cd /cluster/data/droSec1 mkdir chunksUnsplit faSplit about scaffolds.fa 100000 chunksUnsplit/chunk_ ssh kkr1u00 mkdir /iscratch/i/droSec1 rsync -av /cluster/data/droSec1/chunksUnsplit /iscratch/i/droSec1/ rsync -av /cluster/data/droSec1/droSec1.2bit /iscratch/i/droSec1/ # Use -repMatch=100 (based on size -- for human we use 1024, and # fly size is ~4.4% of human judging by gapless dm1 genome size from # featureBits -- we would use 45, but bump that up a bit to be more # conservative). mkdir /cluster/bluearc/droSec1 blat /cluster/data/droSec1/droSec1.2bit /dev/null /dev/null -tileSize=11 \ -makeOoc=/cluster/bluearc/droSec1/11.ooc -repMatch=100 #Wrote 7825 overused 11-mers to /cluster/bluearc/droSec1/11.ooc cp -p /cluster/bluearc/droSec1/*.ooc /iscratch/i/droSec1/ iSync # PRODUCING GENSCAN PREDICTIONS (DONE 11/30/05 angie) ssh kkstore02 # Make hard-masked scaffolds and split up for processing: cd /cluster/data/droSec1 maskOutFa scaffolds.fa hard scaffolds.fa.masked mkdir chunksUnsplitMasked faSplit about scaffolds.fa.masked 100000 chunksUnsplitMasked/chunk_ mkdir /cluster/data/droSec1/bed/genscan cd /cluster/data/droSec1/bed/genscan # Check out hg3rdParty/genscanlinux to get latest genscan: cvs co hg3rdParty/genscanlinux # Make 3 subdirectories for genscan to put their output files in mkdir gtf pep subopt ls -1S ../../chunksUnsplitMasked/chunk*.fa > chunks.list cat << '_EOF_' > gsub #LOOP /cluster/bin/x86_64/gsBig {check in line+ $(path1)} {check out line gtf/$(root1).gtf} -trans={check out line pep/$(root1).pep} -subopt={check out line subopt/$(root1).bed} -exe=hg3rdParty/genscanlinux/genscan -par=hg3rdParty/genscanlinux/HumanIso.smat -tmp=/tmp -window=2400000 #ENDLOOP '_EOF_' # << this line keeps emacs coloring happy ssh kki cd /cluster/data/droSec1/bed/genscan gensub2 chunks.list single gsub jobList para make jobList para time #Completed: 514 of 514 jobs #Average job time: 13s 0.22m 0.00h 0.00d #Longest finished job: 674s 11.23m 0.19h 0.01d #Submission to last job: 674s 11.23m 0.19h 0.01d # If there are crashes, diagnose with "para problems". # If a job crashes due to genscan running out of memory, re-run it # manually with "-window=1200000" instead of "-window=2400000". # Concatenate scaffold-level results: ssh kkstore02 cd /cluster/data/droSec1/bed/genscan cat gtf/*.gtf > genscan.gtf cat subopt/*.bed > genscanSubopt.bed cat pep/*.pep > genscan.pep # Clean up rm -r /cluster/data/droSec1/chunksUnsplitMasked # Load into the database as so: ssh hgwdev cd /cluster/data/droSec1/bed/genscan ldHgGene -gtf droSec1 genscan genscan.gtf hgPepPred droSec1 generic genscanPep genscan.pep hgLoadBed droSec1 genscanSubopt genscanSubopt.bed # MAKE DOWNLOADABLE FILES (DONE 11/30/05 angie) ssh kolossus mkdir /cluster/data/droSec1/bigZips cd /cluster/data/droSec1 gzip -c RMOut/scaffolds.fa.out > bigZips/scaffoldOut.gz gzip -c scaffolds.fa > bigZips/scaffoldFa.gz gzip -c scaffolds.fa.masked > bigZips/scaffoldFaMasked.gz gzip -c bed/simpleRepeat/trfMask.bed > bigZips/scaffoldTrf.gz cd bigZips md5sum *.gz > md5sum.txt # create a README.txt ssh hgwdev mkdir /usr/local/apache/htdocs/goldenPath/droSec1 cd /usr/local/apache/htdocs/goldenPath/droSec1 mkdir bigZips database # Create README.txt files in database/ to explain the files. ln -s /cluster/data/droSec1/bigZips/*.{gz,txt} bigZips # SWAP CHAINS FROM DM2, BUILD NETS ETC. (DONE 12/5/05 angie) mkdir /cluster/data/droSec1/bed/blastz.dm2.swap cd /cluster/data/droSec1/bed/blastz.dm2.swap doBlastzChainNet.pl -swap /cluster/data/dm2/bed/blastz.droSec1/DEF \ >& do.log echo "check /cluster/data/droSec1/bed/blastz.dm2.swap/do.log" \ | mail -s "check do.log" $USER # Add {chain,net}Dm2 to trackDb.ra if necessary. # AUTO UPDATE GENBANK INITIAL RUN (DONE 12/6/05 angie) # align with revised genbank process cd ~kent/src/makeDb/genbank cvs update -d etc # edit etc/genbank.conf to add droSec1 # droSec1 (D. sechellia) # genbank.150.0/full has 7 mRNAs and 0 ESTs at time of initial build, # so no need to make native tracks. droSec1.serverGenome = /cluster/data/droSec1/droSec1.2bit droSec1.clusterGenome = /iscratch/i/droSec1/droSec1.2bit droSec1.ooc = /iscratch/i/droSec1/11.ooc droSec1.lift = no droSec1.refseq.mrna.native.pslCDnaFilter = ${lowCover.refseq.mrna.native.pslCDnaFilter} droSec1.refseq.mrna.xeno.pslCDnaFilter = ${lowCover.refseq.mrna.xeno.pslCDnaFilter} droSec1.genbank.mrna.native.pslCDnaFilter = ${lowCover.genbank.mrna.native.pslCDnaFilter} droSec1.genbank.mrna.xeno.pslCDnaFilter = ${lowCover.genbank.mrna.xeno.pslCDnaFilter} droSec1.genbank.est.native.pslCDnaFilter = ${lowCover.genbank.est.native.pslCDnaFilter} droSec1.genbank.est.xeno.pslCDnaFilter = ${lowCover.genbank.est.xeno.pslCDnaFilter} droSec1.refseq.mrna.native.load = no droSec1.refseq.mrna.xeno.load = yes droSec1.genbank.mrna.native.load = no droSec1.genbank.mrna.xeno.load = yes droSec1.genbank.est.native.load = no droSec1.genbank.est.xeno.load = no droSec1.downloadDir = droSec1 droSec1.perChromTables = no # update /cluster/data/genbank/ cvs ci -m "Added droSec1." etc/genbank.conf make etc-update # droSec1 is a new species, so add it to src/lib/gbGenome.c make cvs ci -m "Added droSec." src/lib/gbGenome.c make install-server ssh kkstore02 cd /cluster/data/genbank nice bin/gbAlignStep -initial droSec1 & # when finished ssh hgwdev cd /cluster/data/genbank nice ./bin/gbDbLoadStep -drop -initialLoad droSec1& featureBits droSec1 xenoRefGene #27947465 bases of 157238575 (17.774%) in intersection featureBits droSec1 xenoMrna #24574087 bases of 157238575 (15.629%) in intersection # MAKE Drosophila Proteins track (DONE braney 2005-12-07) ssh kkstore02 mkdir -p /cluster/data/droSec1/blastDb cd /cluster/data/droSec1/blastDb faSplit sequence ../scaffolds.fa 400 x for i in *.fa; do /cluster/bluearc/blast229/formatdb -i $i -p F 2> /dev/null; done rm *.fa *.log cd .. ssh kk destDir=/cluster/panasas/home/store/droSec1/blastDb mkdir -p $destDir cp /cluster/data/droSec1/blastDb/* $destDir mkdir -p /cluster/data/droSec1/bed/tblastn.dm2FB cd /cluster/data/droSec1/bed/tblastn.dm2FB ls -1S $destDir/*.nsq | sed "s/\.nsq//" > target.lst mkdir fbfa # calculate a reasonable number of jobs calc `wc /cluster/data/dm2/bed/blat.dm2FB/dm2FB.psl|awk "{print \\\$1}"`/\(80000/`wc target.lst | awk "{print \\\$1}"`\) # 18929/(80000/396) = 93.698550 split -l 94 /cluster/data/dm2/bed/blat.dm2FB/dm2FB.psl fbfa/fb cd fbfa for i in *; do pslxToFa $i $i.fa; rm $i; done cd .. ls -1S fbfa/*.fa > fb.lst mkdir -p /cluster/bluearc/droSec1/bed/tblastn.dm2FB/blastOut ln -s /cluster/bluearc/droSec1/bed/tblastn.dm2FB/blastOut for i in `cat fb.lst`; do mkdir blastOut/`basename $i .fa`; done tcsh cat << '_EOF_' > blastGsub #LOOP blastSome $(path1) {check in line $(path2)} {check out exists blastOut/$(root2)/q.$(root1).psl } #ENDLOOP '_EOF_' cat << '_EOF_' > blastSome #!/bin/sh BLASTMAT=/iscratch/i/blast/data export BLASTMAT g=`basename $2` f=/tmp/`basename $3`.$g for eVal in 0.01 0.001 0.0001 0.00001 0.000001 1E-09 1E-11 do if /scratch/blast/blastall -M BLOSUM80 -m 0 -F no -e $eVal -p tblastn -d $1 -i $2 -o $f.8 then mv $f.8 $f.1 break; fi done if test -f $f.1 then if /cluster/bin/i386/blastToPsl $f.1 $f.2 then liftUp -nosort -type=".psl" -pslQ -nohead $3.tmp /cluster/data/dm2/bed/blat.dm2FB/protein.lft warn $f.2 mv $3.tmp $3 rm -f $f.1 $f.2 $f.3 $f.4 exit 0 fi fi rm -f $f.1 $f.2 $3.tmp $f.3 $f.8 $f.4 exit 1 '_EOF_' chmod +x blastSome gensub2 target.lst fb.lst blastGsub blastSpec para create blastSpec para push # Completed: 79992 of 79992 jobs # CPU time in finished jobs: 2936163s 48936.05m 815.60h 33.98d 0.093 y # IO & Wait Time: 653523s 10892.05m 181.53h 7.56d 0.021 y # Average job time: 45s 0.75m 0.01h 0.00d # Longest running job: 0s 0.00m 0.00h 0.00d # Longest finished job: 13727s 228.78m 3.81h 0.16d # Submission to last job: 25997s 433.28m 7.22h 0.30d ssh kkstore02 cd /cluster/data/droSec1/bed/tblastn.dm2FB tcsh cat << '_EOF_' > chainGsub #LOOP chainSome $(path1) #ENDLOOP '_EOF_' cat << '_EOF_' > chainSome (cd $1; cat q.*.psl | /cluster/bin/i386/simpleChain -prot -outPsl -maxGap=25000 stdin ../c.`basename $1`.psl) '_EOF_' chmod +x chainSome ls -1dS `pwd`/blastOut/fb?? > chain.lst gensub2 chain.lst single chainGsub chainSpec exit # out of cshell ssh kki cd /cluster/data/droSec1/bed/tblastn.dm2FB para create chainSpec # Completed: 202 of 202 jobs # CPU time in finished jobs: 337s 5.61m 0.09h 0.00d 0.000 y # IO & Wait Time: 4606s 76.77m 1.28h 0.05d 0.000 y # Average job time: 24s 0.41m 0.01h 0.00d # Longest finished job: 80s 1.33m 0.02h 0.00d # Submission to last job: 430s 7.17m 0.12h 0.00d cd /cluster/data/droSec1/bed/tblastn.dm2FB/blastOut for i in fb?? do awk "(\$13 - \$12)/\$11 > 0.6 {print}" c.$i.psl > c60.$i.psl sort -rn c60.$i.psl | pslUniq stdin u.$i.psl awk "((\$1 / \$11) ) > 0.60 { print }" c60.$i.psl > m60.$i.psl echo $i done sort -T /tmp -k 14,14 -k 16,16n -k 17,17n u.*.psl m60* | uniq > /cluster/data/droSec1/bed/tblastn.dm2FB/blastDm2FB.psl cd .. wc blastDm2FB.psl # 21575 453075 3313821 blastDm2FB.psl pslUniq blastDm2FB.psl stdout | wc # 18827 395367 2992653 cat fbfa/*fa | grep ">" | wc # 82338 82338 1300520 ssh hgwdev cd /cluster/data/droSec1/bed/tblastn.dm2FB hgLoadPsl droSec1 blastDm2FB.psl featureBits droSec1 blastDm2FB # 20986178 bases of 157238575 (13.347%) in intersection exit # back to kkstore02 rm -rf blastOut # End tblastn ########################################################################### # SWAP/CHAIN/NET DM3 (DONE 6/8/07 angie) ssh kkstore02 mkdir /cluster/data/droSec1/bed/blastz.dm3.swap cd /cluster/data/droSec1/bed/blastz.dm3.swap doBlastzChainNet.pl -swap /cluster/data/dm3/bed/blastz.droSec1/DEF >& do.log & tail -f do.log ln -s blastz.dm3.swap /cluster/data/droSec1/bed/blastz.dm3