cat <(cut -f 1,2 samples.tsv) Human.SRA_curated_predict.meta.tsv | perl -ne 'chomp; $f=$_; @f=split(/\t/,$f); if(scalar(@f) == 2) { $h{$f[1]}=$f[0]; next; } $rid=$h{$f[0]}; $temp=$f[2]; $f[2]=$f[1]; $f[1]=$temp; print "$rid\t".join("\t",@f)."\n";'

cat samples.tsv.studies | perl -ne 'chomp; print "/bin/bash -x study_metadata_layout.sh $_ human sra Human.SRA_curated_predict.meta.w_rids.tsv > runs/$_.run 2>&1\n";' > study_metadata_create.jobs

cat samples.tsv.studies | perl -ne 'chomp; print "/bin/bash -x study_metadata_layout.sh $_ human sra Human.SRA_curated_predict.meta.w_rids.rtypes.tsv > runs/$_.run 2>&1\n";' > study_metadata_create.jobs2

cat <(pcat sra.recount_project.gz | head -1) <(fgrep -f samples.tsv.studies samples.tsv | cut -f 1,2,3 | perl -ne 'chomp; $f=$_; ($rid,$run,$study)=split(/\t/,$f); print "$f\t$study\tHomo sapiens\tdata_sources/sra\tdata_sources/sra\t2019-10-01\n";') | gzip > sra.recount_project.gz.1

cat srav3h.seqtk.w_study.tsv.cut samples.tsv | perl -ne 'chomp; $f=$_; @f=split(/\t/,$f,-1); if(scalar(@f) == 15) { $rid=shift(@f); $run=shift(@f); $study=shift(@f); $f=join("\t",@f); $h{$rid}=$f; next; } $seqtk=$h{$f[0]}; print "$f\t$seqtk\n";' | sort -t'	' -k3,3 > samples.w_seqtk.sorted_by_study.tsv

cut -f 1 ../prep/samples.tsv.studies | perl -ne 'chomp; $s=$_; print "/bin/bash -x ../prep/study_metadata_layout.sh $s mouse sra ../prep/Mouse.SRA_curated_predict.meta.w_rids.w_cell_types_nas.tsv > runs/$s.run 2>&1\n";' > metadata.jobs

cat errs.names R109.noCDS.all_gene_ids.fixed_chrms.with_genes.gtf.genes2.sorted sra.gene_sums.SRP051599.R109.gz.genes | perl -ne 'chomp; $f=$_; @f=split(/\t/,$f,-1); if(scalar(@f) == 2) {$h2{$f[0]}=1; next; } if(scalar(@f) > 2) { $f[8]=~/gene_id "([^\"]+)"/; $g=$1; if($h2{$g}) { $g.=".".$f[0]; } $f=~s/gene_id "([^"]+)"/gene_id "$g"/;  $h{$g}=$f; next; } $line=$h{$f[0]}; if(!$line) { print STDERR "$f\n"; next; } print "$line\n";' 2>errs | gzip > human.gene_sums.R109.gtf.gz2

for f in `cat ../prep/recount_project.files`; do pcat $f | fgrep -v "rail_id" >> all_recount_projects ; done
cat <(pcat ./63/SRP170963/sra.recount_project.SRP170963.MD.gz |head -1) all_recount_projects | gzip > sra.recount_project.MD.gz

cat kevin.updated.bulk.samples.tsv Mouse.SRA_curated_predict.meta.w_rids.w_cell_types_nas.tsv | perl -ne 'chomp; $f=$_; @f=split(/\t/,$f,-1); if(scalar(@f) == 7) { $h{$f[0]}=[$f[2],$f[3],$f[4],$f[5]]; next; } ($study,$ct,$cl,$t)=@{$h{$f[1]}}; if(!$study) { $f[7]="scrna-seq"; } elsif($f[2] ne $study) { print STDERR "studies dont match ".$f[2]." vs $study\n"; } $f[5]=$t; $f[7]="rna-seq"; $f[8]=$ct; $f[9]=$cl; print join("\t",@f)."\n";' > Mouse.SRA_curated_predict.meta.w_rids.w_cell_types_nas.tsv.updated

cat samples.tsv.studies | perl -ne 'chomp; $s=$_; print "/bin/bash -x ../prep/rerelease_preds.sh $s sra ../prep/Mouse.SRA_curated_predict.meta.w_rids.w_cell_types_nas.tsv > runs2/$s.run 2>&1\n";' > redo_preds.jobs

cat samples.tsv.studies | perl -ne 'chomp; print "/bin/bash -x ./rerelease_preds.sh $_ sra Human.SRA_curated_predict.meta.w_rids.tsv > runs/$_.run 2>&1\n";' > study_metadata_prediction_release.jobs