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

#########################################################################
Thu Apr  9 07:52:21 PDT 2020 Max

cd /hive/data/genomes/wuhCor1/bed/primers
# got primers.tsv from https://docs.google.com/spreadsheets/d/17eucK_mSAGgYvxVCKJkbbYaogr2ofbhgW2f9W7K3oOA/edit#gid=636736390
python convert.py > primers.fa 
blat ../../wuhCor1.2bit primers.fa stdout -stepSize=3 -tileSize=6  -minScore=10 -oneOff=1 -noHead -fine | pslReps stdin stdout /dev/null -minNearTopSize=10 -minCover=0.8 -nohead > primers.psl 
hgLoadPsl wuhCor1 primers.psl 
hgLoadSeq wuhCor1 primers.fa -replace
cut -f10 primers.psl  | sort > found.txt


#########################################################################
# SWIFT primers (DONE 2021-03-10 angie)

mkdir /hive/data/genomes/wuhCor1/bed/swift
cd /hive/data/genomes/wuhCor1/bed/swift
wget https://swiftbiosci.com/wp-content/uploads/2020/09/sarscov2_v1_masterfile.txt.zip
unzip sarscov2_v1_masterfile.txt.zip
head sarscov2_v1_masterfile.txt
#NC_045512.2     200     301     covid19genome_200-29703_s7490_U_88      182     200     covid19genome_200-29703_s7490_U_88F     301     322     covid19genome_200-29703_s7490_U_88R     CTGCAGGCTGCTTACGGT      AGGCAAACTGAGTTGGACGTG
#...
# Looks like they use interbase coords, good, columns appear to be
# chrom, amplicon{Start,End,Name}, forward{Start,End,Name}, reverse{Start,End,Name}, {forw,rev}Seq
# I don't see any info about pools etc on https://swiftbiosci.com/swift-amplicon-sars-cov-2-panel/
# so BED 6 + sequence should do here.
perl -wne 'chomp; chomp;
           (undef, undef, undef, undef, $fSt, $fE, $fN, $rSt, $rE, $rN, $fSeq, $rSeq) = split;
           print join("\t", "NC_045512v2", $fSt, $fE, $fN, 0, "+", $fSeq) . "\n";
           print join("\t", "NC_045512v2", $rSt, $rE, $rN, 0, "-", $rSeq) . "\n";
    ' sarscov2_v1_masterfile.txt \
| sed -re 's/covid19genome_200-29703_//' \
| sort -k1,1 -k2n,2n > swift.bed
cat > swift.as <<EOF
table swift
"Browser extensible data (6 fields) plus sequence "
    (
    string chrom;      "Chromosome (or contig, scaffold, etc.)"
    uint   chromStart; "Start position"
    uint   chromEnd;   "End position"
    string name;       "Name"
    uint   score;      "Score from 0-1000 (BED placeholder)"
    char[1] strand;    "+ (forward) or - (reverse)"
    string sequence;   "Sequence"
    )
EOF
bedToBigBed -type=bed6+1 -as=swift.as swift.bed ../../chrom.sizes swift.bb
# Link into bbi along with other primer tracks (artic.bb, rapid.bb):
ln -s `pwd`/swift.bb /gbdb/wuhCor1/bbi/


#########################################################################
# ARTIC V4 (DONE 2021-07-22 angie)

mkdir /hive/data/genomes/wuhCor1/bed/artic/v4
cd /hive/data/genomes/wuhCor1/bed/artic/v4
wget https://raw.githubusercontent.com/artic-network/artic-ncov2019/master/primer_schemes/nCoV-2019/V4/SARS-CoV-2.primer.bed
tawk '{ print "NC_045512v2", $2, $3, $4, 0, $6, $5; }' SARS-CoV-2.primer.bed \
| sort -k2n,2n \
    > articV4.bed
bedToBigBed -type=bed6+1 -as=../artic.as -tab articV4.bed ../../../chrom.sizes articV4.bb
ln -s `pwd`/articV4.bb /gbdb/wuhCor1/bbi/


#########################################################################
# ARTIC V4.1 (DONE 2022-01-10 angie)

mkdir /hive/data/genomes/wuhCor1/bed/artic/v4.1
cd /hive/data/genomes/wuhCor1/bed/artic/v4.1
wget https://raw.githubusercontent.com/artic-network/artic-ncov2019/master/primer_schemes/nCoV-2019/V4.1/SARS-CoV-2.primer.bed
tawk '{ print "NC_045512v2", $2, $3, $4, 0, $6, $5; }' SARS-CoV-2.primer.bed \
| sort -k2n,2n \
    > articV4.1.bed
bedToBigBed -type=bed6+1 -as=../artic.as -tab articV4.1.bed ../../../chrom.sizes articV4.1.bb
ln -s `pwd`/articV4.1.bb /gbdb/wuhCor1/bbi/


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