## allocation of CDS sequences to subgenomes

# assigned to subgenome D
zcat Bhybridum.ABR113.primaryTrs.cds.fna.gz | grep -P "hy.D\d+" | wc
  37711  188555 4751586
zcat BhybridumBd28.v1.primaryTrs.cds.fna.gz | grep -P "28.D\d+" | wc
  37641  188205 4893330
zcat Bhybridumv30.primaryTrs.cds.fna.gz | grep -P "30.D\d+" | wc
  35599  177995 4912662
zcat BhybridumBhyb26.v1.primaryTrs.cds.fna.gz | grep -P "26.D\d+" | wc
  34267  171335 4591778
zcat Bhybridumv118.5.primaryTrs.cds.fna.gz | grep -P "118_5.D\d+" | wc
  26920  134600 4038000

# assigned to subgenome S
zcat Bhybridum.ABR113.primaryTrs.cds.fna.gz | grep -P "hy.S\d+" | wc
  32449  162245 4088574
zcat BhybridumBd28.v1.primaryTrs.cds.fna.gz | grep -P "28.S\d+" | wc
  32024  160120 4163120
zcat Bhybridumv30.primaryTrs.cds.fna.gz | grep -P "30.S\d+" | wc
  30941  154705 4269858
zcat BhybridumBhyb26.v1.primaryTrs.cds.fna.gz | grep -P "26.S\d+" | wc
  29219  146095 3915346
zcat Bhybridumv118.5.primaryTrs.cds.fna.gz | grep -P "118_5.S\d+" | wc
  26533  132665 3979950

# summary table
	primary_transcripts	in_subgenomes	%	ambiguous	%	subgenomeD	%	subgenomeS	%
ABR113		70160	70160	100	0	0	37711	53.75	32449		46.25
Bd28			70383	69665	98.98	718	1.02	37641	54.03	32024	45.97
Bhyb30		67960	66540	97.91	1420	2.09	35599	53.5	30941	46.5
Bhyb26		66977	63486	94.79	3491	5.21	34267	53.98	29219	46.02
Bhyb118-5	67466	53453	79.23	14013	20.77	26920	50.36	26533	49.64
									
BdistachyonBd213v1	36647													
Bstaceiv0.1	36357								

# subgenome extraction to FASTA files
zcat Bhybridum.ABR113.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ pr
int "$s\n$fa{$s}" if($s =~ m/hy.D\d+/) }}' > Bhybridum.ABR113.subgenomeD.cds.faa
zcat Bhybridum.ABR113.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ pr
int "$s\n$fa{$s}" if($s =~ m/hy.D\d+/) }}' > Bhybridum.ABR113.subgenomeD.cds.fna
zcat Bhybridum.ABR113.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ pr
int "$s\n$fa{$s}" if($s =~ m/hy.S\d+/) }}' > Bhybridum.ABR113.subgenomeS.cds.fna
zcat Bhybridum.ABR113.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ pr
int "$s\n$fa{$s}" if($s =~ m/hy.S\d+/) }}' > Bhybridum.ABR113.subgenomeS.cds.faa

zcat BhybridumBd28.v1.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ pr
int "$s\n$fa{$s}" if($s =~ m/28.D\d+/) }}' > BhybridumBd28.v1.subgenomeD.cds.faa
zcat BhybridumBd28.v1.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ pr
int "$s\n$fa{$s}" if($s =~ m/28.D\d+/) }}' > BhybridumBd28.v1.subgenomeD.cds.fna
zcat BhybridumBd28.v1.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ pr
int "$s\n$fa{$s}" if($s =~ m/28.S\d+/) }}' > BhybridumBd28.v1.subgenomeS.cds.fna
zcat BhybridumBd28.v1.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ pr
int "$s\n$fa{$s}" if($s =~ m/28.S\d+/) }}' > BhybridumBd28.v1.subgenomeS.cds.faa

zcat Bhybridumv30.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ print 
"$s\n$fa{$s}" if($s =~ m/30.D\d+/) }}' > Bhybridumv30.subgenomeD.cds.faa
zcat Bhybridumv30.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ print 
"$s\n$fa{$s}" if($s =~ m/30.D\d+/) }}' > Bhybridumv30.subgenomeD.cds.fna
zcat Bhybridumv30.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ print 
"$s\n$fa{$s}" if($s =~ m/30.S\d+/) }}' > Bhybridumv30.subgenomeS.cds.fna
zcat Bhybridumv30.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ print 
"$s\n$fa{$s}" if($s =~ m/30.S\d+/) }}' > Bhybridumv30.subgenomeS.cds.faa

zcat BhybridumBhyb26.v1.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ 
print "$s\n$fa{$s}" if($s =~ m/26.D\d+/) }}' > BhybridumBhyb26.v1.subgenomeD.cds.faa
zcat BhybridumBhyb26.v1.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ 
print "$s\n$fa{$s}" if($s =~ m/26.D\d+/) }}' > BhybridumBhyb26.v1.subgenomeD.cds.fna
zcat BhybridumBhyb26.v1.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ 
print "$s\n$fa{$s}" if($s =~ m/26.S\d+/) }}' > BhybridumBhyb26.v1.subgenomeS.cds.fna
zcat BhybridumBhyb26.v1.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ 
print "$s\n$fa{$s}" if($s =~ m/26.S\d+/) }}' > BhybridumBhyb26.v1.subgenomeS.cds.faa

zcat Bhybridumv118.5.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ print "$s\n$fa{$s}" if($s =~ m/118_5.D\d+/) }}' > Bhyb118-5.subgenomeD.cds.faa
zcat Bhybridumv118.5.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ print "$s\n$fa{$s}" if($s =~ m/118_5.D\d+/) }}' > Bhyb118-5.subgenomeD.cds.fna
zcat Bhybridumv118.5.primaryTrs.cds.faa.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ print "$s\n$fa{$s}" if($s =~ m/118_5.S\d+/) }}' > Bhyb118-5.subgenomeS.cds.faa
eeat Bhybridumv118.5.primaryTrs.cds.fna.gz | perl -lne 'if(/^(>.*)/){ $head=$1 } else { $fa{$head} .= $_ } END{ foreach $s (sort(keys(%fa))){ print "$s\n$fa{$s}" if($s =~ m/118_5.S\d+/) }}' > Bhyb118-5.subgenomeS.cds.fna

gzip *subgenome*f?a