#############################################################################
## 5-Way Multiz (DONE - 2016-12-17 - Hiram)
    ssh hgwdev
    mkdir /hive/data/genomes/hg38/bed/multiz5way
    cd /hive/data/genomes/hg38/bed/multiz5way

    # from the 218-way in the source tree, select out the 5 used here:
    /cluster/bin/phast/tree_doctor \
        --prune-all-but hg38,galVar1,cavPor3,mm10,tupChi1 \
        /cluster/home/hiram/kent/src/hg/utils/phyloTrees/218way.nh \
          > hg38.5way.nh
    cat hg38.5way.nh
    # ((hg38:0.143908,(tupChi1:0.120000,galVar1:0.080000):0.054937):0.002000,
    #	(mm10:0.315424,cavPor3:0.175779):0.041059);

    # no need for rearrangement since hg38 is already at the top

    #	what that looks like:
 ~/kent/src/hg/utils/phyloTrees/asciiTree.pl hg38.5way.nh | sed -e 's/^/# /;'
# ((hg38:0.143908,
#  (tupChi1:0.12,
#  galVar1:0.080000):0.054937):0.002000,
# (mm10:0.315424,
# cavPor3:0.175779):0.041059);


    # extract species list from that .nh file
    sed 's/[a-z][a-z]*_//g; s/:[0-9\.][0-9\.]*//g; s/;//; /^ *$/d' \
        hg38.5way.nh | xargs echo | sed 's/ //g; s/,/ /g' \
        | sed 's/[()]//g; s/,/ /g' | tr '[ ]' '[\n]' > species.list.txt

    # construct db to name translation list:
    cat species.list.txt | while read DB
do
hgsql -N -e "select name,organism from dbDb where name=\"${DB}\";" hgcentraltest
done | sed -e "s/\t/->/; s/ /_/g;" | sed -e 's/$/;/' | sed -e 's/\./_/g' \
        | sed -e 's/-nosed/_nosed/; s/-eating/_eating/;' > db.to.name.txt

    # construct a common name .nh file:
    /cluster/bin/phast/tree_doctor --rename \
    "`cat db.to.name.txt`" hg38.5way.nh | sed -e 's/00*)/)/g; s/00*,/,/g' \
       | $HOME/kent/src/hg/utils/phyloTrees/asciiTree.pl /dev/stdin \
         > hg38.5way.commonNames.nh
    cat hg38.5way.commonNames.nh | sed -e 's/^/# /;'
# ((Human:0.143908,
#  (Chinese_tree_shrew:0.12,
#  Malayan_flying_lemur:0.08):0.054937):0.002,
# (Mouse:0.315424,
# Guinea_pig:0.175779):0.041059);

#	Use this specification in the phyloGif tool:
#	http://genome.ucsc.edu/cgi-bin/phyloGif
#	to obtain a png image for src/hg/htdocs/images/phylo/hg38_5way.png

    ~/kent/src/hg/utils/phyloTrees/asciiTree.pl hg38.5way.nh > t.nh
    ~/kent/src/hg/utils/phyloTrees/scientificNames.sh t.nh \
       | $HOME/kent/src/hg/utils/phyloTrees/asciiTree.pl /dev/stdin \
          > hg38.5way.scientificNames.nh
    rm -f t.nh
    cat hg38.5way.scientificNames.nh | sed -e 's/^/# /;'
# ((Homo_sapiens:0.143908,
#  (Tupaia_chinensis:0.12,
#  Galeopterus_variegatus:0.08):0.054937):0.002,
# (Mus_musculus:0.315424,
# Cavia_porcellus:0.175779):0.041059);

    /cluster/bin/phast/all_dists hg38.5way.nh | grep hg38 \
        | sed -e "s/hg38.//" | sort -k2n > 5way.distances.txt
    #	Use this output to create the table below
    cat 5way.distances.txt | sed -e 's/^/# /;'
# galVar1       0.278845
# tupChi1       0.318845
# cavPor3       0.362746
# mm10  0.502391

    printf '#!/usr/bin/env perl

use strict;
use warnings;

open (FH, "<5way.distances.txt") or
        die "can not read 5way.distances.txt";

my $count = 0;
while (my $line = <FH>) {
    chomp $line;
    my ($D, $dist) = split('"'"'\\s+'"'"', $line);
    my $chain = "chain" . ucfirst($D);
    my $B="/hive/data/genomes/hg38/bed/lastz.$D/fb.hg38." .
        $chain . "Link.txt";
    my $chainLinkMeasure =
        `awk '"'"'{print \\$5}'"'"' ${B} 2> /dev/null | sed -e "s/(//; s/)//"`;
    chomp $chainLinkMeasure;
    $chainLinkMeasure = 0.0 if (length($chainLinkMeasure) < 1);
    $chainLinkMeasure =~ s/\\%%//;
    my $swapFile="/hive/data/genomes/${D}/bed/lastz.hg38/fb.${D}.chainHg38Link.txt";
    my $swapMeasure = "N/A";
    if ( -s $swapFile ) {
	$swapMeasure =
	    `awk '"'"'{print \\$5}'"'"' ${swapFile} 2> /dev/null | sed -e "s/(//; s/)//"`;
	chomp $swapMeasure;
	$swapMeasure = 0.0 if (length($swapMeasure) < 1);
	$swapMeasure =~ s/\\%%//;
    }
    my $orgName=
    `hgsql -N -e "select organism from dbDb where name='"'\$D'"';" hgcentraltest`;
    chomp $orgName;
    if (length($orgName) < 1) {
        $orgName="N/A";
    }
    ++$count;
    printf "# %%02d  %%.4f (%%%% %%05.3f) (%%%% %%05.3f) - %%s %%s\\n", $count, $dist,
        $chainLinkMeasure, $swapMeasure, $orgName, $D;
}
close (FH);
' > sizeStats.pl
    chmod +x ./sizeStats.pl
    ./sizeStats.pl

#	If you can fill in all the numbers in this table, you are ready for
#	the multiple alignment procedure

#       featureBits chainLink measures
#               chainLink
#  N distance  on hg38  on other     other species
# 01  0.2788 (% 57.036) (% 64.881) - Malayan flying lemur galVar1
# 02  0.3188 (% 45.256) (% 50.350) - Chinese tree shrew tupChi1
# 03  0.3627 (% 42.371) (% 48.000) - Guinea pig cavPor3
# 04  0.5024 (% 31.653) (% 35.372) - Mouse mm10

# None of this concern for distances matters in building the first step, the
# maf files.  The distances will be better calibrated later.

    # create species list and stripped down tree for autoMZ
    sed 's/[a-z][a-z]*_//g; s/:[0-9\.][0-9\.]*//g; s/;//; /^ *$/d' \
	hg38.5way.nh | xargs echo | sed 's/ //g; s/,/ /g' > tree.nh

    sed 's/[()]//g; s/,/ /g' tree.nh > species.list
    # hg38 tupChi1 galVar1 mm10 cavPor3


    # survey N50 for each
    for db in `cat species.list`
do
n50.pl /hive/data/genomes/$db/chrom.sizes
done
#       reading: /hive/data/genomes/hg38/chrom.sizes
#       contig count: 455, total size: 3209286105, one half size: 1604643052
# cumulative    N50 count       contig  contig size
1547391171      8       chrX    156040895
1604643052 one half size
1692529807      9       chr8    145138636
#       reading: /hive/data/genomes/tupChi1/chrom.sizes
#       contig count: 50750, total size: 2846580235, one half size: 1423290117
# cumulative    N50 count       contig  contig size
1419920836      231     KB321095        3691413
1423290117 one half size
1423590960      232     KB321106        3670124
#       reading: /hive/data/genomes/galVar1/chrom.sizes
#       contig count: 179514, total size: 3187660572, one half size: 1593830286
# cumulative    N50 count       contig  contig size
1593691350      3422    NW_007730159v1  245222
1593830286 one half size
1593936539      3423    NW_007729331v1  245189
#       reading: /hive/data/genomes/mm10/chrom.sizes
#       contig count: 66, total size: 2730871774, one half size: 1365435887
# cumulative    N50 count       contig  contig size
1312176979      8       chr7    145441459
1365435887 one half size
1442871972      9       chr10   130694993
#       reading: /hive/data/genomes/cavPor3/chrom.sizes
#       contig count: 3144, total size: 2723219641, one half size: 1361609820
# cumulative    N50 count       contig  contig size
1356838683      27      scaffold_25     28222655
1361609820 one half size
1384780737      28      scaffold_27     27942054

    #	bash shell syntax here ...
    cd /hive/data/genomes/hg38/bed/multiz5way
    export H=/hive/data/genomes/hg38/bed
    mkdir mafLinks
    # good assemblies can use syntenic net:
    #  galVar1 mm10 cavPor3 tupChi1
    for G in galVar1 mm10 cavPor3 tupChi1
    do
      mkdir mafLinks/$G
      echo ln -s ${H}/lastz.$G/axtChain/hg38.${G}.synNet.maf.gz ./mafLinks/$G
      ln -s ${H}/lastz.$G/axtChain/hg38.${G}.synNet.maf.gz ./mafLinks/$G
    done

    # other assemblies using recip best net:

    # verify the symLinks are good:
    ls -ogrtL mafLinks/*/* | sed -e 's/^/# /; s/-rw-rw-r-- 1//;'
#   999907313 Feb 19  2015 mafLinks/tupChi1/hg38.tupChi1.synNet.maf.gz
#   710111073 Apr  9  2015 mafLinks/mm10/hg38.mm10.synNet.maf.gz
#   939260340 Apr 28  2015 mafLinks/cavPor3/hg38.cavPor3.synNet.maf.gz
#  1144172783 Apr 26  2016 mafLinks/galVar1/hg38.galVar1.synNet.maf.gz

    # split the maf files into a set of hashed named files
    # this hash named split keeps the same chr/contig names in the same
    # named hash file.
    mkdir /hive/data/genomes/hg38/bed/multiz5way/mafSplit
    cd /hive/data/genomes/hg38/bed/multiz5way/mafSplit
    time for D in `sed -e "s/hg38 //" ../species.list`
do
    echo "${D}"
    mkdir $D
    cd $D
    echo "mafSplit -byTarget -useHashedName=8 /dev/null . ../../mafLinks/${D}/*.maf.gz"
    mafSplit -byTarget -useHashedName=8 /dev/null . \
	../../mafLinks/${D}/*.maf.gz
    cd ..
done
    # real    2m52.152s

    # construct a list of all possible maf file names.
    # they do not all exist in each of the species directories
    find . -type f | wc -l
    # 409
    find . -type f | grep ".maf$" | xargs -L 1 basename | sort -u > maf.list
    wc -l maf.list
    # 108 maf.list

    mkdir /hive/data/genomes/hg38/bed/multiz5way/splitRun
    cd /hive/data/genomes/hg38/bed/multiz5way/splitRun
    mkdir maf run
    cd run
    mkdir penn
    cp -p /cluster/bin/penn/multiz.2009-01-21_patched/multiz penn
    cp -p /cluster/bin/penn/multiz.2009-01-21_patched/maf_project penn
    cp -p /cluster/bin/penn/multiz.2009-01-21_patched/autoMZ penn

    #	verify the db and pairs settings are correct
    printf '#!/bin/csh -ef
set db = hg38
set c = $1
set result = $2
set run = `/bin/pwd`
set tmp = /dev/shm/$db/multiz.$c
set pairs = /hive/data/genomes/hg38/bed/multiz5way/mafSplit
/bin/rm -fr $tmp
/bin/mkdir -p $tmp
/bin/cp -p ../../tree.nh ../../species.list $tmp
pushd $tmp > /dev/null
foreach s (`/bin/sed -e "s/$db //" species.list`)
    set in = $pairs/$s/$c
    set out = $db.$s.sing.maf
    if (-e $in.gz) then
        /bin/zcat $in.gz > $out
        if (! -s $out) then
            echo "##maf version=1 scoring=autoMZ" > $out
        endif
    else if (-e $in) then
        /bin/ln -s $in $out
    else
        echo "##maf version=1 scoring=autoMZ" > $out
    endif
end
set path = ($run/penn $path); rehash
$run/penn/autoMZ + T=$tmp E=$db "`cat tree.nh`" $db.*.sing.maf $c \
        > /dev/null
popd > /dev/null
/bin/rm -f $result
/bin/cp -p $tmp/$c $result
/bin/rm -fr $tmp
' > autoMultiz.csh

    chmod +x autoMultiz.csh

    printf '#LOOP
./autoMultiz.csh $(file1) {check out line+ /hive/data/genomes/hg38/bed/multiz5way/splitRun/maf/$(root1).maf}
#ENDLOOP
' > template

# << happy emacs

    ln -s ../../mafSplit/maf.list maf.list
    ssh ku
    cd /hive/data/genomes/hg38/bed/multiz5way/splitRun/run
    gensub2 maf.list single template jobList
    para create jobList
    para try ... check ... push ... etc...
# Completed: 108 of 108 jobs
# CPU time in finished jobs:      69577s    1159.61m    19.33h    0.81d  0.002 y
# IO & Wait Time:                   409s       6.82m     0.11h    0.00d  0.000 y
# Average job time:                 648s      10.80m     0.18h    0.01d
# Longest finished job:            6028s     100.47m     1.67h    0.07d
# Submission to last job:          6042s     100.70m     1.68h    0.07d

    # combine into one file  (the 1>&2 redirect sends the echo to stderr)
    cd /hive/data/genomes/hg38/bed/multiz5way
    head -1 splitRun/maf/006.maf > multiz5way.maf
    time for F in splitRun/maf/*.maf
do
    echo "${F}" 1>&2
    egrep -v "^#" ${F}
done >> multiz5way.maf
    # real    0m31.596s

    tail -1 splitRun/maf/006.maf >> multiz5way.maf
# -rw-rw-r-- 1 10395239286 Dec 17 21:03 multiz5way.maf

    # Load into database
    ssh hgwdev
    cd /hive/data/genomes/hg38/bed/multiz5way
    mkdir /gbdb/hg38/multiz5way
    ln -s `pwd`/multiz5way.maf /gbdb/hg38/multiz5way
    cd /dev/shm
    time hgLoadMaf hg38 multiz5way
# Loaded 9874234 mafs in 1 files from /gbdb/hg38/multiz5way
# real    2m55.355s

    time hgLoadMafSummary -verbose=2 -minSize=30000 \
	-mergeGap=1500 -maxSize=200000 hg38 multiz5waySummary \
	/gbdb/hg38/multiz5way/multiz5way.maf
# Created 878615 summary blocks from 23435349 components and 9874234 mafs from /gbdb/hg38/multiz5way/multiz5way.maf
# real    3m36.817s

# -rw-rw-r--   1  530340768 Dec 17 21:05 multiz5way.tab
# -rw-rw-r--   1   40949243 Dec 17 21:30 multiz5waySummary.tab

    wc -l multiz5way*.tab
#   9874234 multiz5way.tab
#    878615 multiz5waySummary.tab

    rm multiz5way*.tab

##############################################################################
# GAP ANNOTATE MULTIZ7WAY MAF AND LOAD TABLES (DONE - 2017-12-17 - Hiram)
    # mafAddIRows has to be run on single chromosome maf files, it does not
    #	function correctly when more than one reference sequence
    #	are in a single file.  Need to split of the maf file into individual
    #   maf files
    mkdir -p /hive/data/genomes/hg38/bed/multiz5way/anno/mafSplit
    cd /hive/data/genomes/hg38/bed/multiz5way/anno/mafSplit

    time mafSplit -outDirDepth=2 -byTarget -useFullSequenceName \
        /dev/null . ../../multiz5way.maf
    #   real    2m57.457s
    find . -type f | wc -l
    #   332

    # check for N.bed files everywhere:
    cd /hive/data/genomes/hg38/bed/multiz5way/anno
    for DB in `cat ../species.list`
do
    if [ ! -s /hive/data/genomes/${DB}/${DB}.N.bed ]; then
        echo "MISS: ${DB}"
#         cd /hive/data/genomes/${DB}
#         twoBitInfo -nBed ${DB}.2bit ${DB}.N.bed
    else
        echo "  OK: ${DB}"
    fi
done

    cd /hive/data/genomes/hg38/bed/multiz5way/anno
    for DB in `cat ../species.list`
do
    echo "${DB} "
    ln -s  /hive/data/genomes/${DB}/${DB}.N.bed ${DB}.bed
    echo ${DB}.bed  >> nBeds
    ln -s  /hive/data/genomes/${DB}/chrom.sizes ${DB}.len
    echo ${DB}.len  >> sizes
done
    # make sure they all are successful symLinks:
    ls -ogrtL

    screen -S gapAnno      # use a screen to control this longish job
    ssh ku
    cd /hive/data/genomes/hg38/bed/multiz5way/anno
    mkdir result
    find ./mafSplit -type d | sed -e 's#./mafSplit/##; s#./mafSplit##;' \
      | while read D
do
    echo mkdir -p result/${D}
    mkdir -p result/${D}
done
    printf '#LOOP
mafAddIRows -nBeds=nBeds mafSplit/$(path1) /hive/data/genomes/hg38/hg38.2bit {check out exists+ result/$(path1)}
#ENDLOOP
' > template

    find ./mafSplit -type f | sed -e 's#^./mafSplit/##' > maf.list
    gensub2 maf.list single template jobList
    # limit jobs on a node with the ram=32g requirement because they go fast
    para -ram=32g create jobList
    para try ... check ... push ...
# Completed: 332 of 332 jobs
# CPU time in finished jobs:        717s      11.95m     0.20h    0.01d  0.000 y
# IO & Wait Time:                   906s      15.10m     0.25h    0.01d  0.000 y
# Average job time:                   5s       0.08m     0.00h    0.00d
# Longest finished job:              62s       1.03m     0.02h    0.00d
# Submission to last job:            96s       1.60m     0.03h    0.00d

    # verify all result files have some content, look for 0 size files:
    find ./result -type f -size 0
    # should see none
    # or in this manner:
    find ./result -type f | xargs ls -og | sort -k3nr | tail

    # combine into one file  (the 1>&2 redirect sends the echo to stderr)
    head -q -n 1 result/0/0/chr14_KI270845v1_alt.maf > hg38.5way.maf
    time find ./result -type f | while read F
do
    echo "${F}" 1>&2
    grep -h -v "^#" ${F}
done >> hg38.5way.maf
    # real    0m43.949s


    #	these maf files do not have the end marker, this does nothing:
    #	tail -q -n 1 result/0/0/chr14_KI270845v1_alt.maf >> hg38.5way.maf
    # How about an official end marker:
    echo "##eof maf" >> hg38.5way.maf
    ls -og
# -rw-rw-r--  1 13290077466 Dec 17 22:37 hg38.5way.maf

    du -hsc hg38.5way.maf ../*.maf
    # 13G     hg38.5way.maf
    #  9.7G    ../multiz5way.maf

    # construct symlinks to get the individual maf files into gbdb:
    rm /gbdb/hg38/multiz5way/multiz5way.maf   # remove previous results
    ln -s `pwd`/hg38.5way.maf /gbdb/hg38/multiz5way/multiz5way.maf

    # Load into database
    cd /dev/shm
    time hgLoadMaf -pathPrefix=/gbdb/hg38/multiz5way hg38 multiz5way
    # Loaded 11023447 mafs in 1 files from /gbdb/hg38/multiz5way
    # real    3m37.392s


    time hgLoadMafSummary -verbose=2 -minSize=30000 \
	-mergeGap=1500 -maxSize=200000 hg38 multiz5waySummary \
        /gbdb/hg38/multiz5way/multiz5way.maf
    # Created 878615 summary blocks from 23435349 components and 11023447 mafs from /gbdb/hg38/multiz5way/multiz5way.maf
    # real    4m10.706s

    # -rw-rw-r--   1  594646492 Dec 17 22:41 multiz5way.tab
    # -rw-rw-r--   1   42706473 Dec 17 22:48 multiz5waySummary.tab

    rm multiz5way*.tab

######################################################################
# MULTIZ7WAY MAF FRAMES (TBD - 2016-05-05 - Hiram)
    ssh hgwdev
    mkdir /hive/data/genomes/hg38/bed/multiz5way/frames
    cd /hive/data/genomes/hg38/bed/multiz5way/frames
#   survey all the genomes to find out what kinds of gene tracks they have
    printf '#!/bin/csh -fe
foreach db (`cat ../species.list`)
    printf "# ${db}: "
    set tables = `hgsql $db -N -e "show tables" | egrep "Gene|ncbiRefSeq"`
    foreach table ($tables)
        if ($table == "ensGene" || $table == "refGene" || \
           $table == "ncbiRefSeq" || $table == "mgcGenes" || \
           $table == "knownGene" || $table == "xenoRefGene" ) then
           set count = `hgsql $db -N -e "select count(*) from $table"`
            echo -n "${table}: ${count}, "
        endif
    end
    set orgName = `hgsql hgcentraltest -N -e \
            "select scientificName from dbDb where name='"'"'$db'"'"'"`
    set orgId = `hgsql hgFixed -N -e \
            "select id from organism where name='"'"'$orgName'"'"'"`
    if ($orgId == "") then
        echo "Mrnas: 0"
    else
        set count = `hgsql hgFixed -N -e "select count(*) from gbCdnaInfo where organism=$orgId"`
        echo "Mrnas: ${count}"
    endif
end
' > showGenes.csh
    # << happy emacs
    chmod +x ./showGenes.csh
    time ./showGenes.csh
# hg38: ensGene: 208239, knownGene: 196838, mgcGenes: 35312, ncbiRefSeq: 159322, refGene: 74453, xenoRefGene: 187376, Mrnas: 11508577
# tupChi1: refGene: 206, xenoRefGene: 353563, Mrnas: 50709
# galVar1: ncbiRefSeq: 41547, xenoRefGene: 516902, Mrnas: 0
# mm10: ensGene: 103734, knownGene: 63759, mgcGenes: 27612, ncbiRefSeq: 106520, refGene: 39240, xenoRefGene: 183459, Mrnas: 5371140
# cavPor3: ensGene: 34846, refGene: 488, xenoRefGene: 316945, Mrnas: 21241

#  real    1m5.600s

    # from that summary, use these gene sets:
    # knownGene - hg38 mm10
    # ensGene - cavPor3
    # none - galVar1 tupChi1
 
    mkdir genes
    #   1. knownGene: hg38 mm10
    for DB in hg38 mm10
do
    hgsql -N -e "select name,chrom,strand,txStart,txEnd,cdsStart,cdsEnd,exonCount,exonStarts,exonEnds from knownGene" ${DB} \
      | genePredSingleCover stdin stdout | gzip -2c \
        > genes/${DB}.gp.gz
    printf "# ${DB}: "
    genePredCheck -db=${DB} genes/${DB}.gp.gz
done
# hg38: checked: 21554 failed: 0
# mm10: checked: 21100 failed: 0

    #   2. ensGene: cavPor3
    for DB in cavPor3
do
hgsql -N -e "select name,chrom,strand,txStart,txEnd,cdsStart,cdsEnd,exonCount,exonStarts,exonEnds
from ensGene" ${DB} \
      | genePredSingleCover stdin stdout | gzip -2c \
        > /dev/shm/${DB}.tmp.gz
    mv /dev/shm/${DB}.tmp.gz genes/$DB.gp.gz
    printf "# ${DB}: "
    genePredCheck -db=${DB} genes/${DB}.gp.gz
done
# cavPor3: checked: 18034 failed: 0

    # verify counts for genes are reasonable:
    for T in genes/*.gz
do
    echo -n "# $T: "
    zcat $T | cut -f1 | sort | uniq -c | wc -l
done
# genes/cavPor3.gp.gz: 18034
# genes/hg38.gp.gz: 21554
# genes/mm10.gp.gz: 21100

    time (cat ../anno/hg38.5way.maf \
	| genePredToMafFrames hg38 stdin stdout \
          `sed -e 's/tupChi1//; s/galVar1//;' ../species.list.txt | xargs echo \
            | sed -e "s#\([a-zA-Z0-9]*\)#\1 genes/\1.gp.gz#g;"` \
		| gzip > multiz5wayFrames.bed.gz)
    # real    2m39.249s

    # verify there are frames on everything, should be 3 species:
    zcat multiz5wayFrames.bed.gz | awk '{print $4}' | sort | uniq -c \
       | sed -e 's/^/# /;'
#  241239 cavPor3
#  206671 hg38
#  261120 mm10

    #   load the resulting file
    ssh hgwdev
    cd /hive/data/genomes/hg38/bed/multiz5way/frames
    time hgLoadMafFrames hg38 multiz5wayFrames multiz5wayFrames.bed.gz
    #   real    0m7.413s

    time featureBits -countGaps hg38 multiz5wayFrames
    # 39254551 bases of 3209286105 (1.223%) in intersection
    # real    0m5.378s

    #   enable the trackDb entries:
# frames multiz5wayFrames
# irows on
    #   appears to work OK

#########################################################################
# Phylogenetic tree from 5-way (TBD - 2016-05-06 - Hiram)
    mkdir /hive/data/genomes/hg38/bed/multiz5way/4d
    cd /hive/data/genomes/hg38/bed/multiz5way/4d

    # using knownGene for hg38, only transcribed genes and nothing
    #	from the randoms and other misc.
    hgsql -Ne "select name,chrom,strand,txStart,txEnd,cdsStart,cdsEnd,exonCount,exonStarts,exonEnds from knownGene where cdsEnd > cdsStart;" hg38 \
      | egrep -E -v "chrM|chrUn|random|_alt" \
        | genePredSingleCover stdin stdout > hg38.knownGeneNR.gp
    genePredCheck -db=hg38 hg38.knownGeneNR.gp
    # checked: 19306 failed: 0

    # the annotated maf is:
    og ../anno/hg38.5way.maf
# -rw-rw-r-- 1 13290077466 Dec 17 22:37 ../anno/hg38.5way.maf

    mkdir annoSplit
    cd annoSplit
    time mafSplit -verbose=2 -outDirDepth=2 -byTarget -useFullSequenceName \
	/dev/null . ../../anno/hg38.5way.maf
    # real    4m9.934s

    find . -type f | wc -l
    #   332
    ssh ku
    mkdir /hive/data/genomes/hg38/bed/multiz5way/4d/run
    cd /hive/data/genomes/hg38/bed/multiz5way/4d/run
    mkdir ../mfa

    cat << '_EOF_' > 4d.csh
#!/bin/csh -fex
set PHASTBIN = /cluster/bin/phast.build/cornellCVS/phast.2010-12-30/bin
set GP = hg38.knownGeneNR.gp

set r = "/hive/data/genomes/hg38/bed/multiz5way"
set c = $1
set infile = $r/4d/$2
set outDir = $r/4d/$3
set outfile = $r/4d/run/$4
/bin/mkdir -p $outDir
cd /dev/shm
/bin/awk -v C=$c '$2 == C {print}' $r/4d/$GP | sed -e "s/\t$c\t/\ttupChi1.$c\t/" > $c.gp
set NL=`wc -l $c.gp| gawk '{print $1}'`
echo $NL
if ("$NL" != "0") then
    $PHASTBIN/msa_view --4d --features $c.gp -i MAF $infile -o SS > $c.ss
    $PHASTBIN/msa_view -i SS --tuple-size 1 $c.ss > $outfile
else
    echo "" > $outfile
endif
/bin/rm -f /dev/shm/$c.gp /dev/shm/$c.ss
_EOF_
    # << happy emacs

    chmod +x 4d.csh

    find ../annoSplit -type f | sed -e "s#../annoSplit/##" > maf.list
    wc -l maf.list
# 332 maf.list

    printf '#LOOP
4d.csh $(root1) annoSplit/$(dir1)/$(file1) mfa/$(dir1) {check out line+ ../mfa/$(dir1)/$(root1).mfa}
#ENDLOOP
' > template

    gensub2 maf.list single template jobList
    para create jobList
    para try ... check
    para time
# Completed: 330 of 332 jobs
# Crashed: 2 jobs
# CPU time in finished jobs:        585s       9.76m     0.16h    0.01d  0.000 y
# IO & Wait Time:                   773s      12.88m     0.21h    0.01d  0.000 y
# Average job time:                   4s       0.07m     0.00h    0.00d
# Longest finished job:              54s       0.90m     0.01h    0.00d
# Submission to last job:            91s       1.52m     0.03h    0.00d

    # Not all results have contents, or finish successfully, that is OK
    # it is because not all contigs have genes, only gene sequences are measured

    # combine mfa files
    ssh hgwdev
    cd /hive/data/genomes/hg38/bed/multiz5way/4d
    # remove the broken empty files, size 0 and size 1:
    find ./mfa -type f -size 0 | xargs rm -f
    # sometimes this doesn't work, don't know why
    ### XXX does not work find ./mfa -type f -size 1 | xargs rm -f
    # when it doesn't, use this empty list procedure
    find ./mfa -type f | xargs ls -og | awk '$3 < 2' | awk '{print $NF}' \
        > empty.list
    cat empty.list | xargs rm -f
    # see what is left:
    ls -ogrt mfa/*/*/*.mfa | sort -k3nr | wc
    #  22     154    1091

    # want comma-less species.list
    time /cluster/bin/phast.build/cornellCVS/phast.2010-12-30/bin/msa_view \
	--aggregate "`cat ../species.list`" mfa/*/*/*.mfa | sed s/"> "/">"/ \
	    > 4d.all.mfa
    # real    0m0.426s

    # check they are all in there:
    grep "^>" 4d.all.mfa | wc -l
    # 5
    grep "^>" 4d.all.mfa | sed -e 's/^/# /;'
# >hg38
# >tupChi1
# >galVar1
# >mm10
# >cavPor3

    sed 's/[a-z][a-z]*_//g; s/:[0-9\.][0-9\.]*//g; s/;//; /^ *$/d' \
	../hg38.5way.nh | xargs echo | sed -e 's/ //g' > tree_commas.nh
    # tree_commas.nh looks like:
    # ((hg38,(tupChi1,galVar1)),(mm10,cavPor3))

    # use phyloFit to create tree model (output is phyloFit.mod)
    time /cluster/bin/phast.build/cornellCVS/phast.2010-12-30/bin/phyloFit \
	    --EM --precision MED --msa-format FASTA --subst-mod REV \
		--tree tree_commas.nh 4d.all.mfa
    #  real    0m0.256s

    mv phyloFit.mod all.mod

    grep TREE all.mod
# TREE:
# ((hg38:0.0700315,(tupChi1:0.0981977,galVar1:0.0629139):0.00881734):0.0161669,
#  (mm10:0.142138,cavPor3:0.112136):0.0161669);

    # compare these calculated lengths to the tree extracted from 218way:
    grep TREE all.mod | sed -e 's/TREE: //' \
       | /cluster/bin/phast/all_dists /dev/stdin | grep hg38 \
          | sed -e "s/hg38.//;"  | sort > new.dists
    /cluster/bin/phast/all_dists ../hg38.5way.nh | grep hg38 \
        | sed -e "s/hg38.//;" | sort > old.dists
     # printing out the 'new', the 'old' the 'difference' and percent difference
    join new.dists old.dists | awk '{
  printf "#\t%s\t%8.6f\t%8.6f\t%8.6f\t%8.6f\n", $1, $2, $3, $2-$3, 100*($2-$3)/$3 }' \
      | sort -k3n
#       galVar1 0.141763        0.278845        -0.137082       -49.160645
#       tupChi1 0.177047        0.318845        -0.141798       -44.472393
#       cavPor3 0.214501        0.362746        -0.148245       -40.867439
#       mm10    0.244503        0.502391        -0.257888       -51.332130

XXX - ready to continue - Mon Dec 18 11:34:03 PST 2017
#########################################################################
# phastCons 5-way (TBD - 2016-05-06 - Hiram)
    # split 5way mafs into 10M chunks and generate sufficient statistics
    # files for # phastCons
    ssh ku
    mkdir -p /hive/data/genomes/hg38/bed/multiz5way/cons/SS
    cd /hive/data/genomes/hg38/bed/multiz5way/cons/SS
    mkdir result done

    printf '#!/bin/csh -ef
set d = $1
set c = $2
set doneDir = done/$d
set MAF = /hive/data/genomes/hg38/bed/multiz5way/anno/result/$d/$c.maf
set WINDOWS = /hive/data/genomes/hg38/bed/multiz5way/cons/SS/result/$d/$c
set WC = `cat $MAF | wc -l`
set NL = `grep "^#" $MAF | wc -l`
if ( -s $3 ) then
    exit 0
endif
if ( -s $3.running ) then
    exit 0
endif

/bin/mkdir -p $doneDir
/bin/date >> $3.running

/bin/rm -fr $WINDOWS
/bin/mkdir -p $WINDOWS
pushd $WINDOWS > /dev/null
if ( $WC != $NL ) then
/cluster/bin/phast.build/cornellCVS/phast.2010-12-30/bin/msa_split \\
    $MAF -i MAF -o SS -r $WINDOWS/$c -w 10000000,0 -I 1000 -B 5000
endif
popd > /dev/null
/bin/date >> $3
/bin/rm -f $3.running
' > mkSS.csh

    chmod +x mkSS.csh

    printf '#LOOP
mkSS.csh $(dir1) $(root1) {check out line+ done/$(dir1)/$(root1)}
#ENDLOOP
' > template

    find ../../anno/result -type f | sed -e "s#../../anno/result/##" > maf.list
    wc -l maf.list
# 76237 maf.list

    ssh ku
    cd /hive/data/genomes/hg38/bed/multiz5way/cons/SS

    gensub2 maf.list single template jobList
    # beware overwhelming the cluster with these quick high I/O jobs
    para create jobList
    para try ... check ... etc
    para -maxJob=64 push
# Completed: 76237 of 76237 jobs
# CPU time in finished jobs:       3491s      58.19m     0.97h    0.04d  0.000 y
# IO & Wait Time:                321255s    5354.25m    89.24h    3.72d  0.010 y
# Average job time:                   4s       0.07m     0.00h    0.00d
# Longest finished job:              10s       0.17m     0.00h    0.00d
# Submission to last job:          1583s      26.38m     0.44h    0.02d


    find ./result -type f | wc -l
    # 24853

    # Run phastCons
    #	This job is I/O intensive in its output files, beware where this
    #	takes place or do not run too many at once.
    ssh ku
    mkdir -p /hive/data/genomes/hg38/bed/multiz5way/cons/run.cons
    cd /hive/data/genomes/hg38/bed/multiz5way/cons/run.cons

    #	This is setup for multiple runs based on subsets, but only running
    #   the 'all' subset here.
    #   It triggers off of the current working directory
    #	$cwd:t which is the "grp" in this script.  Running:
    #	all and vertebrates

    printf '#!/bin/csh -fe
set PHASTBIN = /cluster/bin/phast.build/cornellCVS/phast.2010-12-30/bin
set c = $1
set d = $2
set f = $3
set len = $4
set cov = $5
set rho = $6
set grp = $cwd:t
set cons = /hive/data/genomes/hg38/bed/multiz5way/cons
set tmp = $cons/tmp/${d}_${c}
mkdir -p $tmp
set ssSrc = $cons/SS/result
set useGrp = "$grp.mod"
if (-s $cons/$grp/$grp.non-inf) then
  ln -s $cons/$grp/$grp.mod $tmp
  ln -s $cons/$grp/$grp.non-inf $tmp
  ln -s $ssSrc/$d/$f $tmp
else
  ln -s $ssSrc/$d/$f $tmp
  ln -s $cons/$grp/$grp.mod $tmp
endif
pushd $tmp > /dev/null
if (-s $grp.non-inf) then
  $PHASTBIN/phastCons $f $useGrp \
    --rho $rho --expected-length $len --target-coverage $cov --quiet \\
    --not-informative `cat $grp.non-inf` \\
    --seqname $c --idpref $c --most-conserved $c.bed --score > $c.pp
else
  $PHASTBIN/phastCons $f $useGrp \\
    --rho $rho --expected-length $len --target-coverage $cov --quiet \\
    --seqname $c --idpref $c --most-conserved $c.bed --score > $c.pp
endif
popd > /dev/null
mkdir -p pp/$d bed/$d
sleep 4
touch pp/$d bed/$d
rm -f pp/$d/$c.pp
rm -f bed/$d/$c.bed
mv $tmp/$c.pp pp/$d
mv $tmp/$c.bed bed/$d
rm -fr $tmp
rmdir --ignore-fail-on-non-empty $cons/tmp/$d:h
' > doPhast.csh

    chmod +x doPhast.csh

    #	this template will serve for all runs
    #	root1 == chrom name, file1 == ss file name without .ss suffix
    printf '#LOOP
../run.cons/doPhast.csh $(root1) $(dir1) $(file1) 45 0.3 0.3 {check out line+ pp/$(dir1)/$(root1).pp}
#ENDLOOP
' > template

    find ../SS/result -type f | sed -e "s#../SS/result/##" > ss.list
    wc -l ss.list
    #	24853 ss.list

    # Create parasol batch and run it
    # run for all species
    cd /hive/data/genomes/hg38/bed/multiz5way/cons
    mkdir -p all
    cd all
    #	Using the .mod tree
    cp -p ../../4d/all.mod ./all.mod

    gensub2 ../run.cons/ss.list single ../run.cons/template jobList
    para -ram=32g create jobList
    para try ... check ...
    para push
# Completed: 24815 of 24853 jobs
# Crashed: 38 jobs
# CPU time in finished jobs:       6246s     104.09m     1.73h    0.07d  0.000 y
# IO & Wait Time:                222621s    3710.36m    61.84h    2.58d  0.007 y
# Average job time:                   9s       0.15m     0.00h    0.00d
# Longest finished job:              18s       0.30m     0.01h    0.00d
# Submission to last job:           935s      15.58m     0.26h    0.01d

   # the 38 crash jobs were actually finished, they failed the last rmdir:
# rmdir: failed to remove `/hive/data/genomes/hg38/bed/multiz5way/cons/tmp/7/6': No such file or directory

    # create Most Conserved track
    cd /hive/data/genomes/hg38/bed/multiz5way/cons/all
    time cut -f1 ../../../../chrom.sizes | while read C
do
    ls -d bed/?/?/${C} 2> /dev/null | while read D
    do
        echo ${D}/${C}*.bed 1>&2
        cat ${D}/${C}*.bed
    done | sort -k1,1 -k2,2n \
    | awk '{printf "%s\t%d\t%d\tlod=%d\t%s\n", "'${C}'", $2, $3, $5, $5;}'
done > tmpMostConserved.bed
    # real    19m26.845s

    time /cluster/bin/scripts/lodToBedScore tmpMostConserved.bed \
         > mostConserved.bed
    # real    0m5.667s

    # -rw-rw-r--  1 36626033 May  6 14:55 mostConserved.bed

    # load into database
    ssh hgwdev
    cd /hive/data/genomes/hg38/bed/multiz5way/cons/all
    time hgLoadBed hg38 phastConsElements5way mostConserved.bed
    # Read 932856 elements of size 5 from mostConserved.bed
    # real    0m9.898s

    # on human we often try for 5% overall cov, and 70% CDS cov
    # most bets are off here for that goal, these alignments are too few
    #	and too far between
    #	--rho 0.3 --expected-length 45 --target-coverage 0.3
    time featureBits hg38 -enrichment ncbiRefSeq:cds phastConsElements5way
# ncbiRefSeq:cds 1.217%, phastConsElements5way 3.976%, both 0.819%,
#   cover 67.27%, enrich 16.92x
    # real    2m33.330s

    # Create merged posterier probability file and wiggle track data files
    cd /hive/data/genomes/hg38/bed/multiz5way/cons/all
    mkdir downloads

    # the third sed fixes the chrom names, removing the partition extensions
    time (find ./pp -type f | sed -e "s#^./##; s#\.# d #g; s#-# m #;" \
	| sort -k1,1 -k3,3n | sed -e "s# d #.#g; s# m #-#g;" | xargs cat \
	| sed -e 's/\.[0-9][0-9]*-[0-9][0-9]* start/ start/' \
        | gzip -c > downloads/phastCons5way.wigFix.gz)
    #   real    27m47.145s
# -rw-rw-r-- 1 2207346680 May  6 15:33 phastCons5way.wigFix.gz

    # check integrity of data with wigToBigWig
    time (zcat downloads/phastCons5way.wigFix.gz \
	| wigToBigWig -verbose=2 stdin /hive/data/genomes/hg38/chrom.sizes \
	    phastCons5way.bw) > bigWig.log 2>&1
    egrep "real|VmPeak" bigWig.log
    # pid=37326: VmPeak:    20943944 kB
    # real    30m30.283s

    bigWigInfo phastCons5way.bw | sed -e 's/^/# /;'
# version: 4
# isCompressed: yes
# isSwapped: 0
# primaryDataSize: 3,364,371,979
# primaryIndexSize: 81,959,008
# zoomLevels: 10
# chromCount: 24853
# basesCovered: 1,910,088,593
# mean: 0.117235
# min: 0.000000
# max: 1.000000
# std: 0.237776

    #	encode those files into wiggle data
    time (zcat downloads/phastCons5way.wigFix.gz \
	| wigEncode stdin phastCons5way.wig phastCons5way.wib)
    # Converted stdin, upper limit 1.00, lower limit 0.00
    #  real    10m31.797s

    du -hsc *.wi?
    #  1.8G    phastCons5way.wib
    #  275M    phastCons5way.wig

    # Load gbdb and database with wiggle.
    ln -s `pwd`/phastCons5way.wib /gbdb/hg38/multiz5way/phastCons5way.wib
    time hgLoadWiggle -pathPrefix=/gbdb/hg38/multiz5way \
	hg38 phastCons5way phastCons5way.wig
    #   real    0m30.803s

    # use to set trackDb.ra entries for wiggle min and max
    # and verify table is loaded correctly

    wigTableStats.sh hg38 phastCons5way
# db.table               min max mean       count sumData
# hg38.phastCons5way   0 1 0.117235 1910088593 2.23929e+08
#     stdDev  viewLimits
#    0.237776 viewLimits=0:1

    #  Create histogram to get an overview of all the data
    time hgWiggle -doHistogram -db=hg38 \
	-hBinSize=0.001 -hBinCount=1000 -hMinVal=0.0 -verbose=2 \
	    phastCons5way > histogram.data 2>&1
    #	real    4m6.489s

    #	create plot of histogram:

    printf 'set terminal png small x000000 xffffff xc000ff x66ff66 xffff00 x00ffff font \
"/usr/share/fonts/default/Type1/n022004l.pfb"
set size 1.4, 0.8
set key left box
set grid noxtics
set grid ytics
set title " Malayan flying lemur hg38 Histogram phastCons5way track"
set xlabel " phastCons5way score"
set ylabel " Relative Frequency"
set y2label " Cumulative Relative Frequency (CRF)"
set y2range [0:1]
set y2tics
set yrange [0:0.02]

plot "histogram.data" using 2:5 title " RelFreq" with impulses, \\
        "histogram.data" using 2:7 axes x1y2 title " CRF" with lines
' | gnuplot > histo.png

    display histo.png &

#########################################################################
# phyloP for 5-way (TBD - 2016-05-09,11 - Hiram)
    # run phyloP with score=LRT
    ssh ku
    mkdir /cluster/data/hg38/bed/multiz5way/consPhyloP
    cd /cluster/data/hg38/bed/multiz5way/consPhyloP

    mkdir run.phyloP
    cd run.phyloP
    # Adjust model file base composition background and rate matrix to be
    # representative of the chromosomes in play
    grep BACKGROUND ../../4d/all.mod | awk '{printf "%0.3f\n", $3 + $4}'
    #	0.562
    /cluster/bin/phast.build/cornellCVS/phast.2010-12-30/bin/modFreqs \
	../../4d/all.mod 0.562 > all.mod
    # verify, the BACKGROUND should now be paired up:
    grep BACK all.mod
    #   BACKGROUND: 0.219000 0.281000 0.281000 0.219000 

    printf '#!/bin/csh -fe
set PHASTBIN = /cluster/bin/phast.build/cornellCVS/phast.2010-12-30/bin
set f = $1
set d = $f:h
set file1 = $f:t
set out = $2
set cName = $f:t:r
set grp = $cwd:t
set cons = /hive/data/genomes/hg38/bed/multiz5way/consPhyloP
set tmp = $cons/tmp/$grp/$f
/bin/rm -fr $tmp
/bin/mkdir -p $tmp
set ssSrc = "/hive/data/genomes/hg38/bed/multiz5way/cons/SS/result/$f"
set useGrp = "$grp.mod"
/bin/ln -s $cons/run.phyloP/$grp.mod $tmp
pushd $tmp > /dev/null
$PHASTBIN/phyloP --method LRT --mode CONACC --wig-scores --chrom $cName \\
    -i SS $useGrp $ssSrc.ss > $file1.wigFix
popd > /dev/null
/bin/mkdir -p $out:h
sleep 4
/bin/touch $out:h
/bin/mv $tmp/$file1.wigFix $out
/bin/rm -fr $tmp
/bin/rmdir --ignore-fail-on-non-empty $cons/tmp/$grp/$d
/bin/rmdir --ignore-fail-on-non-empty $cons/tmp/$grp/$d:h
/bin/rmdir --ignore-fail-on-non-empty $cons/tmp/$grp
/bin/rmdir --ignore-fail-on-non-empty $cons/tmp
' > doPhyloP.csh

    chmod +x doPhyloP.csh

    # Create list of chunks
    find ../../cons/SS/result -type f | grep ".ss$" \
	| sed -e "s/.ss$//; s#^../../cons/SS/result/##" > ss.list
    # make sure the list looks good
    wc -l ss.list
    #	24853 ss.list

    # Create template file
    #	file1 == $chr/$chunk/file name without .ss suffix
    printf '#LOOP
../run.phyloP/doPhyloP.csh $(path1) {check out line+ wigFix/$(dir1)/$(file1).wigFix}
#ENDLOOP
' > template

    ######################   Running all species  #######################
    # setup run for all species
    mkdir /hive/data/genomes/hg38/bed/multiz5way/consPhyloP/all
    cd /hive/data/genomes/hg38/bed/multiz5way/consPhyloP/all
    rm -fr wigFix
    mkdir wigFix

    gensub2 ../run.phyloP/ss.list single ../run.phyloP/template jobList
    #	beware overwhelming the cluster with these fast running high I/O jobs
    para create jobList
    para try ... check ... push ... etc ...
    para -maxJob=53 push
    para time > run.time
# Completed: 24852 of 24853 jobs
# Crashed: 1 jobs
# CPU time in finished jobs:       7517s     125.29m     2.09h    0.09d  0.000 y
# IO & Wait Time:                166287s    2771.45m    46.19h    1.92d  0.005 y
# Average job time:                   7s       0.12m     0.00h    0.00d
# Longest finished job:              11s       0.18m     0.00h    0.00d
# Submission to last job:          1799s      29.98m     0.50h    0.02d

    # the one failed job was just the last rmdir command:
# /bin/rmdir: failed to remove `/hive/data/genomes/hg38/bed/multiz5way/consPhyloP/tmp/all/5/2

    mkdir downloads

    time (find ./wigFix -type f | sed -e "s#^./##; s#\.# d #g; s#-# m #;" \
	| sort -k1,1 -k3,3n | sed -e "s# d #.#g; s# m #-#g;" | xargs cat \
	| gzip -c > downloads/phyloP5way.wigFix.gz)
    #   real    30m20.572s

    # check integrity of data with wigToBigWig
    time (zcat downloads/phyloP5way.wigFix.gz \
	| wigToBigWig -verbose=2 stdin /hive/data/genomes/hg38/chrom.sizes \
	phyloP5way.bw) > bigWig.log 2>&1
    egrep "real|VmPeak" bigWig.log
    # pid=19896: VmPeak:    20943916 kB
    # real    215m36.588s

    bigWigInfo phyloP5way.bw | sed -e 's/^/# /;'
# version: 4
# isCompressed: yes
# isSwapped: 0
# primaryDataSize: 3,067,676,886
# primaryIndexSize: 81,959,008
# zoomLevels: 10
# chromCount: 24853
# basesCovered: 1,910,088,593
# mean: 0.061208
# min: -2.599000
# max: 0.833000
# std: 0.611659

    #	encode those files into wiggle data
    time (zcat downloads/phyloP5way.wigFix.gz \
	| wigEncode stdin phyloP5way.wig phyloP5way.wib)
    # Converted stdin, upper limit 0.83, lower limit -2.60
    #    real    9m4.543s

    du -hsc *.wi?
    # 1.8G    phyloP5way.wib
    # 279M    phyloP5way.wig

    # Load gbdb and database with wiggle.
    ln -s `pwd`/phyloP5way.wib /gbdb/hg38/multiz5way/phyloP5way.wib
    time hgLoadWiggle -pathPrefix=/gbdb/hg38/multiz5way hg38 \
	phyloP5way phyloP5way.wig
    # real    0m30.869s

    # use to set trackDb.ra entries for wiggle min and max
    # and verify table is loaded correctly

    wigTableStats.sh hg38 phyloP5way
# db.table      min max mean count sumData
# hg38.phyloP5way  -2.599 0.833 0.0612084 1910088593 1.16914e+08
#       stdDev viewLimits
#     0.611659 viewLimits=-2.599:0.833

    #	that range is: 0.833+2.599 = 3.432 for hBinSize=0.003432

    #  Create histogram to get an overview of all the data
    time hgWiggle -doHistogram \
	-hBinSize=0.003432 -hBinCount=1000 -hMinVal=-2.559 -verbose=2 \
	    -db=hg38 phyloP5way > histogram.data 2>&1
    # real    4m20.444s

    # find the Y range for the 2:5 graph
    grep -v chrom histogram.data | grep "^[0-9]" | ave -col=5 stdin \
      | sed -e 's/^/# /;'
# Q1 0.000087
# median 0.000381
# Q3 0.001361
# average 0.001112
# min 0.000000
# max 0.032088
# count 899
# total 0.999990
# standard deviation 0.002165

    # find the X range for the 2:5 graph
    grep "^[0-9]" histogram.data | ave -col=2 stdin \
      | sed -e 's/^/# /;'
# Q1 -1.773070
# median -1.004300
# Q3 -0.228672
# average -0.962349
# min -2.559000
# max 0.831816
# count 899
# total -865.151612
# standard deviation 0.954068

    #	create plot of histogram:
    printf 'set terminal png small x000000 xffffff xc000ff x66ff66 xffff00 x00ffff font \
"/usr/share/fonts/default/Type1/n022004l.pfb"
set size 1.4, 0.8
set key left box
set grid noxtics
set grid ytics
set title " Guinea pig hg38 Histogram phyloP5way track"
set xlabel " phyloP5way score"
set ylabel " Relative Frequency"
set y2label " Cumulative Relative Frequency (CRF)"
set y2range [0:1]
set y2tics
set xtics
set xrange [-2.6:0.85]
set yrange [0:0.033]

plot "histogram.data" using 2:5 title " RelFreq" with impulses, \
        "histogram.data" using 2:7 axes x1y2 title " CRF" with lines
' | gnuplot > histo.png

    display histo.png &
    # appears to have an odd hole in the data near X=0 ?

#############################################################################
# hgPal downloads (TBD - 2016-05-09,11 - Hiram)
#   FASTA from 5-way for knownGene, refGene and knownCanonical

    ssh hgwdev
    screen -S hg38HgPal
    mkdir /hive/data/genomes/hg38/bed/multiz5way/pal
    cd /hive/data/genomes/hg38/bed/multiz5way/pal
    cat ../species.list | tr '[ ]' '[\n]' > order.list

    # this for loop takes about 2.5 hours on this large count contig assembly
    export mz=multiz5way
    export gp=ncbiRefSeq
    export db=hg38
    export I=0
    export D=0
    mkdir exonAA exonNuc
    printf '#!/bin/sh\n' > $gp.jobs

    time for C in `sort -nk2 ../../../chrom.sizes | cut -f1`
    do
        I=`echo $I | awk '{print $1+1}'`
        D=`echo $D | awk '{print $1+1}'`
        dNum=`echo $D | awk '{printf "%03d", int($1/1000)}'`
        mkdir -p exonNuc/${dNum} > /dev/null
        mkdir -p exonAA/${dNum} > /dev/null
	echo "mafGene -chrom=$C -exons -noTrans $db $mz $gp order.list stdout | gzip -c > exonNuc/${dNum}/$C.exonNuc.fa.gz &"
	echo "mafGene -chrom=$C -exons $db $mz $gp order.list stdout | gzip -c > exonAA/${dNum}/$C.exonAA.fa.gz &"
        if [ $I -gt 16 ]; then
            echo "date"
            echo "wait"
            I=0
        fi
    done >> $gp.jobs
    # real    115m16.333s


    echo "date" >> $gp.jobs
    echo "wait" >> $gp.jobs

    chmod +x  ncbiRefSeq.jobs 

    time (./$gp.jobs) > $gp.jobs.log 2>&1 &
    # real    14m50.750s

    export mz=multiz5way
    export gp=ncbiRefSeq
    time find ./exonAA -type f | grep exonAA.fa.gz | xargs zcat \
     | gzip -c > $gp.$mz.exonAA.fa.gz
    #  real    4m20.026s

    time find ./exonNuc -type f | grep exonNuc.fa.gz | xargs zcat \
     | gzip -c > $gp.$mz.exonNuc.fa.gz
    #   real    4m45.761s

  #  -rw-rw-r--   1 36201970 May 11 11:25 ncbiRefSeq.multiz5way.exonAA.fa.gz
  #  -rw-rw-r--   1 59404213 May 11 11:30 ncbiRefSeq.multiz5way.exonNuc.fa.gz

    export mz=multiz5way
    export gp=ncbiRefSeq
    export db=hg38
    export pd=/usr/local/apache/htdocs-hgdownload/goldenPath/$db/$mz/alignments
    mkdir -p $pd
    md5sum *.fa.gz > md5sum.txt
    ln -s `pwd`/$gp.$mz.exonAA.fa.gz $pd/$gp.exonAA.fa.gz
    ln -s `pwd`/$gp.$mz.exonNuc.fa.gz $pd/$gp.exonNuc.fa.gz
    ln -s `pwd`/md5sum.txt $pd/

    rm -rf exonAA exonNuc

#############################################################################
# construct download files for 5-way (TBD - 2016-05-11 - Hiram)
    mkdir /usr/local/apache/htdocs-hgdownload/goldenPath/hg38/multiz5way
    mkdir /usr/local/apache/htdocs-hgdownload/goldenPath/hg38/phastCons5way
    mkdir /usr/local/apache/htdocs-hgdownload/goldenPath/hg38/phyloP5way
    mkdir /hive/data/genomes/hg38/bed/multiz5way/downloads
    cd /hive/data/genomes/hg38/bed/multiz5way/downloads
    mkdir multiz5way phastCons5way phyloP5way
    cd multiz5way
    time cp -p ../../anno/hg38.5way.maf .
    #   real    0m20.564s

    # -rw-rw-r-- 1 13290077466 Dec 17 22:37 hg38.5way.maf

    du -hsc *
    #  13G     hg38.5way.maf

    time gzip *.maf
    #   real    43m14.605s

    # -rw-rw-r-- 1 3422259860 Dec 17 22:37 hg38.5way.maf.gz

    du -hsc *.maf.gz
    #  3.2G    hg38.5way.maf.gz

    grep TREE ../../4d/all.mod | awk '{print $NF}' \
      | ~/kent/src/hg/utils/phyloTrees/asciiTree.pl /dev/stdin \
         > hg38.5way.nh
    ~/kent/src/hg/utils/phyloTrees/commonNames.sh hg38.5way.nh \
      | ~/kent/src/hg/utils/phyloTrees/asciiTree.pl /dev/stdin \
         > hg38.5way.commonNames.nh
    ~/kent/src/hg/utils/phyloTrees/scientificNames.sh hg38.5way.nh \
	| $HOME/kent/src/hg/utils/phyloTrees/asciiTree.pl /dev/stdin \
	    > hg38.5way.scientificNames.nh
    time md5sum *.nh *.maf.gz > md5sum.txt
    #   real    0m35.144s

    ln -s `pwd`/* \
        /usr/local/apache/htdocs-hgdownload/goldenPath/hg38/multiz5way

    du -hsc *.maf.gz ../../anno/hg38.5way.maf
    #  3.2G     hg38.5way.maf.gz
    #  13G     ../../anno/hg38.5way.maf

XXX - writing README - Mon Dec 18 11:41:15 PST 2017
    # obtain the README.txt from cavPor3/multiz5way and update for this
    #   situation

    #####################################################################
    cd /hive/data/genomes/hg38/bed/multiz5way/downloads/phastCons5way

    ln -s ../../cons/all/downloads/phastCons5way.wigFix.gz \
        ./hg38.phastCons5way.wigFix.gz
    ln -s ../../cons/all/phastCons5way.bw ./hg38.phastCons5way.bw
    ln -s ../../cons/all/all.mod ./hg38.phastCons5way.mod
    time md5sum *.gz *.mod *.bw > md5sum.txt
    #   real    0m20.354s

    # obtain the README.txt from cavPor3/phastCons5way and update for this
    #   situation
    ln -s `pwd`/* \
      /usr/local/apache/htdocs-hgdownload/goldenPath/hg38/phastCons5way

    #####################################################################
    cd /hive/data/genomes/hg38/bed/multiz5way/downloads/phyloP5way

    ln -s ../../consPhyloP/all/downloads/phyloP5way.wigFix.gz \
        ./hg38.phyloP5way.wigFix.gz
    ln -s ../../consPhyloP/run.phyloP/all.mod hg38.phyloP5way.mod
    ln -s ../../consPhyloP/all/phyloP5way.bw hg38.phyloP5way.bw

    time md5sum *.mod *.bw *.gz > md5sum.txt
    #   real    0m29.662s

    # obtain the README.txt from hg38/phyloP17way and update for this
    #   situation
    ln -s `pwd`/* \
      /usr/local/apache/htdocs-hgdownload/goldenPath/hg38/phyloP5way

    ###########################################################################
    ## create upstream refGene maf files
    cd /hive/data/genomes/hg38/bed/multiz5way/downloads/multiz5way
    # bash script
#!/bin/sh
export geneTbl="knownGene"
for S in 1000 2000 5000
do
    echo "making upstream${S}.maf"
    featureBits hg38 ${geneTbl}:upstream:${S} -fa=/dev/null -bed=stdout \
        | perl -wpe 's/_up[^\t]+/\t0/' | sort -k1,1 -k2,2n \
        | /cluster/bin/$MACHTYPE/mafFrags hg38 multiz5way \
                stdin stdout \
                -orgs=/hive/data/genomes/hg38/bed/multiz5way/species.list \
        | gzip -c > upstream${S}.${geneTbl}.maf.gz
    echo "done upstream${S}.${geneTbl}.maf.gz"
done
    #   about 20 minutes

    md5sum *.maf.gz *.nh upstream*.gz README.txt >> md5sum.txt

    # some other symlinks were already made above
    # obtain the README.txt from cavPor3/multiz5way and update for this
    #   situation
    ln -s `pwd`/upstream*.gz `pwd`/README.txt \
        /usr/local/apache/htdocs-hgdownload/goldenPath/hg38/multiz5way

#############################################################################
# hgPal downloads (TBD - 2016-05-11 - Hiram)
#   FASTA from 5-way for knownGene, refGene and knownCanonical

    ssh hgwdev
    screen -S hg38HgPal
    mkdir /hive/data/genomes/hg38/bed/multiz5way/pal
    cd /hive/data/genomes/hg38/bed/multiz5way/pal
    cat ../species.list | tr '[ ]' '[\n]' > order.list

    # this for loop takes about 2.5 hours on this large count contig assembly
    export mz=multiz5way
    export gp=xenoRefGene
    export db=hg38
    export I=0
    export D=0
    mkdir exonAA exonNuc
    for C in `sort -nk2 ../../../chrom.sizes | cut -f1`
    do
        I=`echo $I | awk '{print $1+1}'`
        D=`echo $D | awk '{print $1+1}'`
        dNum=`echo $D | awk '{printf "%03d", int($1/1000)}'`
        mkdir -p exonNuc/${dNum} > /dev/null
        mkdir -p exonAA/${dNum} > /dev/null
	echo "mafGene -chrom=$C -exons -noTrans $db $mz $gp order.list stdout | gzip -c > exonNuc/${dNum}/$C.exonNuc.fa.gz &"
	echo "mafGene -chrom=$C -exons $db $mz $gp order.list stdout | gzip -c > exonAA/${dNum}/$C.exonAA.fa.gz &"
        if [ $I -gt 16 ]; then
            echo "date"
            echo "wait"
            I=0
        fi
    done > $gp.jobs
    echo "date" >> $gp.jobs
    echo "wait" >> $gp.jobs

    time sh -x ./$gp.jobs > $gp.jobs.log 2>&1 &
    # real    175m50.376s


    export mz=multiz5way
    export gp=xenoRefGene
    time find ./exonAA -type f | grep exonAA.fa.gz | xargs zcat \
     | gzip -c > $gp.$mz.exonAA.fa.gz
    # real    10m29.500s

    time find ./exonNuc -type f | grep exonNuc.fa.gz | xargs zcat \
     | gzip -c > $gp.$mz.exonNuc.fa.gz
    #   real    15m9.974s

  # -rw-rw-r--   1 611281544 Apr 15 20:37 xenoRefGene.multiz5way.exonAA.fa.gz
  # -rw-rw-r--   1 955571425 Apr 15 21:06 xenoRefGene.multiz5way.exonNuc.fa.gz

    export mz=multiz5way
    export gp=xenoRefGene
    export db=hg38
    export pd=/usr/local/apache/htdocs-hgdownload/goldenPath/$db/$mz/alignments
    mkdir -p $pd
    md5sum *.fa.gz > md5sum.txt
    ln -s `pwd`/$gp.$mz.exonAA.fa.gz $pd/$gp.exonAA.fa.gz
    ln -s `pwd`/$gp.$mz.exonNuc.fa.gz $pd/$gp.exonNuc.fa.gz
    ln -s `pwd`/md5sum.txt $pd/

    rm -rf exonAA exonNuc

#############################################################################
# wiki page for 5-way (DONE - 2017-12-18 - Hiram)
    mkdir /hive/users/hiram/bigWays/hg38.5way
    cd /hive/users/hiram/bigWays
    echo "hg38" > hg38.5way/ordered.list
    awk '{print $1}' /hive/data/genomes/hg38/bed/multiz5way/5way.distances.txt \
       >> hg38.5way/ordered.list

    # sizeStats.sh catches up the cached measurements required for data
    # in the tables.  They are usually already mostly done, only new
    # assemblies will have updates.
    ./sizeStats.sh hg38.5way/ordered.list
    # dbDb.sh constructs hg38.5way/Hg38_5-way_conservation_alignment.html
    # may need to add new assembly references to srcReference.list and
    # urlReference.list
    ./dbDb.sh hg38 5way
    # sizeStats.pl constructs hg38.5way/Hg38_5-way_Genome_size_statistics.html
    # this requires entries in coverage.list for new sequences
    ./sizeStats.pl hg38 5way

    # defCheck.pl constructs Hg38_5-way_conservation_lastz_parameters.html
    ./defCheck.pl hg38 5way

    # this constructs the html pages in hg38.5way/:
# -rw-rw-r--    3270 Dec 18 13:57 Hg38_5-way_conservation_alignment.html
# -rw-rw-r--    4831 Dec 18 13:57 Hg38_5-way_Genome_size_statistics.html
# -rw-rw-r--    3289 Dec 18 13:57 Hg38_5-way_conservation_lastz_parameters.html

    # add those pages to the genomewiki.  Their page names are the
    # names of the .html files without the .html:
#  Hg38_5-way_conservation_alignment
#  Hg38_5-way_Genome_size_statistics
#  Hg38_5-way_conservation_lastz_parameters

    # when you view the first one you enter, it will have links to the
    # missing two.

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