This is an example to phase subgenomes of an allopolyploid complex using WGDI and SubPhaser. Here we use the data of wheat complex (tetraploid–hexaploid reticulate allopolyploidization) as the example. The complex include an allotetraploid (AABB, Triticum turgidum, 2n = 4x = 28) and an allohexaploid (AABBDD, T. aestivum, 2n = 6x = 42). Less than 0.8 million years ago (mya), a hybridization event between AA (T. urartu) and BB (a close relative of Aegilops speltoides) genomes gave rise to the allopolyploid T. turgidum genome (AABB). Subsequently, less than 0.4 mya, emmer wheat (AABB) hybridized with another wild wheat species carrying the D genome (A. tauschii), resulting in the allohexaploid T. aestivum genome (AABBDD). We assume that the diploid progenitors of allopolyploid wheats were either extinct or not sampled during the subgenome phasing process.
Firstly, we need to install the required software (WGDI and SubPhaser) for this example. Here, we just install the software and dependencies via conda.
git clone https://github.com/zhangrengang/SubPhaser
cd SubPhaser
conda env create -f SubPhaser.yaml -n SGphasing
conda activate SGphasing
python setup.py install
conda install -c bioconda wgdi diamond aster phytop newick_utils
- Genomic data (protein sequences in fasta format and gene coordinates in custom gff format) of the allopolyploid complex are required.
- Genomic data of potential diploid progenitors as far as possible are recommended (here is omitted).
- Genomic data of outgroup or ancestral karyotype are required.
- Configure files for WGDI.
Here, we just use the example data [T. aestivum (AABBDD) and T. turgidum (AABB), and the outgroup Hordeum vulgare] prepared in this repo:
git clone https://github.com/zhangrengang/subgenome_phasing_example
cd subgenome_phasing_example
cd wgdi
gunzip *gz
cat *.pep > pep.faa
cat *.cds > cds.fa
Now, all the required input data are present:
$ tree
├── ak.txt # karyotype of the reference Hordeum_vulgare
├── Hordeum_vulgare.fasta
├── Hordeum_vulgare.gff
├── Hordeum_vulgare.lens
├── Triticum_aestivum.fasta
├── Triticum_aestivum.gff
├── Triticum_aestivum-Hordeum_vulgare.conf
├── Triticum_aestivum.lens
├── Triticum_aestivum-Triticum_aestivum.conf
├── Triticum_turgidum.fasta
├── Triticum_turgidum.gff
├── Triticum_turgidum-Hordeum_vulgare.conf
├── Triticum_turgidum.lens
├── Triticum_turgidum-Triticum_aestivum.conf
Blast results are also required for WGDI. Here, we run the pairwise BLAST search using DIAMOND:
diamond blastp -q Triticum_turgidum.pep -d Triticum_aestivum.pep -o Triticum_turgidum-Triticum_aestivum.blast --more-sensitive -p 40 --quiet -e 0.001
diamond blastp -q Triticum_turgidum.pep -d Hordeum_vulgare.pep -o Triticum_turgidum-Hordeum_vulgare.blast --more-sensitive -p 40 --quiet -e 0.001
diamond blastp -q Triticum_aestivum.pep -d Triticum_aestivum.pep -o Triticum_aestivum-Triticum_aestivum.blast --more-sensitive -p 40 --quiet -e 0.001
diamond blastp -q Triticum_aestivum.pep -d Hordeum_vulgare.pep -o Triticum_aestivum-Hordeum_vulgare.blast --more-sensitive -p 40 --quiet -e 0.001
These processes can be speed up by increasing -p or using parallel computation.
These are basic steps:
wgdi -icl Triticum_turgidum-Triticum_aestivum.conf
wgdi -ks Triticum_turgidum-Triticum_aestivum.conf
wgdi -bi Triticum_turgidum-Triticum_aestivum.conf
wgdi -icl Triticum_turgidum-Hordeum_vulgare.conf
wgdi -ks Triticum_turgidum-Hordeum_vulgare.conf
wgdi -bi Triticum_turgidum-Hordeum_vulgare.conf
wgdi -icl Triticum_aestivum-Triticum_aestivum.conf
wgdi -ks Triticum_aestivum-Triticum_aestivum.conf
wgdi -bi Triticum_aestivum-Triticum_aestivum.conf
wgdi -icl Triticum_aestivum-Hordeum_vulgare.conf
wgdi -ks Triticum_aestivum-Hordeum_vulgare.conf
wgdi -bi Triticum_aestivum-Hordeum_vulgare.conf
To show Ks-colored dot plots:
wgdi -bk Triticum_turgidum-Triticum_aestivum.conf
wgdi -bk Triticum_aestivum-Triticum_aestivum.conf
From the resulted dot plots,
we can find that the D subgenome of Triticum_aestivum shows
higher Ks to Triticum_turgidum, while A or B subgenomes of Triticum_aestivum show lower Ks to Triticum_turgidum (Fig. 1 left).
Thus, the D subgenome as a singleton can be phased out.
 |
 |
|---|
Fig. 1. Ks-colored dot plots between Triticum_turgidum and Triticum_aestivum (left) and within Triticum_aestivum (right). The lower Ks indicates the higher similarity, and the lowest inter-genomic Ks (e.g. the red colored in the left panel) indicates orthology.
We hypothesize the A or B subgenome may be closer to the D subgenome. However, there is no such a pattern that Ks(A-D) is higher or lower than Ks(B-D) to distinguish A and B subgenomes (Fig. 1 right).
Alternatively, Ks can be replaced by Orthology index which shows much more clear orthology relationshipes (Fig. 1c). It is highly recommended when Ks patterns are not clear.
Fig. 1c. Orthology index-colored dot plots between Triticum_turgidum and Triticum_aestivum.
The highest inter-genomic Orthology index (i.e. the red colored dots) indicates orthology.
Fisrt, we need to identify orthologous synteny between the outgroup reference and the polyploids, and to visually validate with the Ks-colored dot plots:
wgdi -c Triticum_turgidum-Hordeum_vulgare.conf
wgdi -bk Triticum_turgidum-Hordeum_vulgare.conf
wgdi -c Triticum_aestivum-Hordeum_vulgare.conf
wgdi -bk Triticum_aestivum-Hordeum_vulgare.conf
If there are non-orthologous syntenic blocks in the dot plots (Fig. 2),
we need to adjust the parameters (including homo, multiple and pvalue)
for wgdi -c, and re-run the above commands. Sometimes, we need to delete the
out-paralogous blocks from the blockinfo file manually.
 |
 |
|---|
Fig. 2. Orthologous synteny.
Then, we map the karyotype of polyploids to the reference:
wgdi -km Triticum_turgidum-Hordeum_vulgare.conf
wgdi -km Triticum_aestivum-Hordeum_vulgare.conf
This step generates the karyotype mapping files of the two wheats: Triticum_turgidum.ancestor.txt and Triticum_aestivum.ancestor.txt.
At this stage, we need to manually edit the two files to assign subgenomes.
We have phased the D subgenome based on Ks-based evidence,
so we number the blocks of D subgenome as 3, but have to randomly number those of A or B subgenomes as 1 or 2. Meanwhile,
the assignments of A/B subgenomes of the two wheats are consistant based on the inter-genomic similarity/orthology.
For eaxmple, if we assign 1A as 2, we need to assign 1B as 1, for Triticum_aestivum, based on their synteny;
accordingly, we need to assign 1A as 2 and 1B as 1, for Triticum_turgidum, based on their orthology to Triticum_aestivum:
$ cat Triticum_aestivum.ancestor.txt
1A 1 4359 RoyalBlue 2
1B 1 4736 RoyalBlue 1
1D 1 4487 RoyalBlue 3
2A 1 5840 red 1
2B 1 6152 red 2
2D 1 5885 red 3
3A 1 5237 #99CC00 1
3B 1 5941 #99CC00 2
3D 1 5306 #99CC00 3
4A 1 3027 deepskyblue 1
4A 3028 3593 #339966 1
4A 3594 3907 fuchsia 2
4A 3908 4056 deepskyblue 1
4A 4057 4870 fuchsia 2
4B 1 3878 deepskyblue 2
4D 1 3582 deepskyblue 3
5A 1 4772 #339966 1
5A 4773 5450 deepskyblue 1
5B 1 5574 #339966 2
5D 1 5574 #339966 3
6A 1 4141 #FFCC00 2
6B 1 4627 #FFCC00 1
6D 1 4012 #FFCC00 3
7A 1 5573 fuchsia 1
7B 1 4892 fuchsia 2
7D 1 5419 fuchsia 3
$ cat Triticum_turgidum.ancestor.txt
1A 1 3906 RoyalBlue 2
1B 1 4136 RoyalBlue 1
2A 1 5192 red 1
2B 1 5463 red 2
3A 1 4956 #99CC00 1
3B 1 5832 #99CC00 2
4A 1 2869 deepskyblue 1
4A 2870 3347 #339966 1
4A 3348 3447 deepskyblue 1
4A 3448 4385 fuchsia 2
4B 1 3487 deepskyblue 2
5A 1 4057 #339966 1
5A 4058 4677 deepskyblue 1
5B 1 5037 #339966 2
6A 1 3651 #FFCC00 2
6B 1 4017 #FFCC00 1
7A 1 4880 fuchsia 1
7B 1 4244 fuchsia 2
The fragmented segments are assigned according to the complementarity of segments. For example, the large segment of chr4A-3' is assigned together with chr7B because they are complementary.
Now, we can apply the assignments (wgdi -pc) and generate alignments (wgdi -a):
wgdi -pc Triticum_turgidum-Hordeum_vulgare.conf
wgdi -a Triticum_turgidum-Hordeum_vulgare.conf
wgdi -pc Triticum_aestivum-Hordeum_vulgare.conf
wgdi -a Triticum_aestivum-Hordeum_vulgare.conf
 |
 |
|---|
Fig. 3. Subgenome assignments based on Ks evidence. The same colored dot plots indicate the same subgenome assignments.
We merge the alignments and build chromosome phylogeny to seek the phylogeny-based evidence (wgdi -at):
paste Triticum_turgidum-Hordeum_vulgare.alignment.csv Triticum_aestivum-Hordeum_vulgare.alignment.csv | perl -pe 's/\t[^,]+//g' > merged.alignment.csv
for chr in $(cut -f1 Hordeum_vulgare.lens)
do
awk -v chr=$chr '$1==chr' Hordeum_vulgare.lens > Hordeum_vulgare.$chr.lens
echo "[alignmenttrees]
alignment = merged.alignment.csv
gff = Hordeum_vulgare.gff
lens = Hordeum_vulgare.$chr.lens
dir = Hordeum_vulgare.$chr.tree
sequence_file = pep.faa
trees_file = Hordeum_vulgare.$chr.trees.nwk
align_software = mafft
tree_software = iqtree
model = MFP
trimming = trimal
minimum = 4
delete_detail = true" > Hordeum_vulgare.$chr.conf
wgdi -at Hordeum_vulgare.$chr.conf
astral-pro -i Hordeum_vulgare.$chr.trees.nwk -u 2 -t 8 -o Hordeum_vulgare.$chr.trees.nwk.astral
phytop -pie -cp Hordeum_vulgare.$chr.trees.nwk.astral
nw_topology -I Hordeum_vulgare.$chr.trees.nwk.astral | nw_order - | nw_display - -w 30
done
The following is the topology of these chromosome phylogenies:
chr1-3H:
/-------------------------+ 1 /-------------------------+ 1 /-------------------------+ 1
| | |
| /------------+ 2 | /------+ 2 | /------+ 2
=+ /-----+ =+ /-----+ =+ /-----+
| | \------------+ 4 | /-----+ \------+ 4 | /-----+ \------+ 4
| | | | | | | |
\------+ /------+ 3 | | \------------+ 6 | | \------------+ 6
| /-----+ \------+ \------+
\-----+ \------+ 5 | /------------+ 3 | /------------+ 3
| \-----+ \-----+
\------------+ 6 \------------+ 5 \------------+ 5
chr4-6H:
/-------------------------+ 1 /-------------------------+ 1 /-------------------------+ 1
| | |
| /------+ 2 | /------+ 2 | /------------+ 2
=+ /-----+ =+ /-----+ =+ /-----+
| /-----+ \------+ 4 | /-----+ \------+ 4 | | \------------+ 4
| | | | | | | |
| | \------------+ 6 | | \------------+ 6 \------+ /------+ 3
\------+ \------+ | /-----+
| /------------+ 3 | /------------+ 3 \-----+ \------+ 5
\-----+ \-----+ |
\------------+ 5 \------------+ 5 \------------+ 6
chr7H:
/-------------------------+ 1
|
| /------+ 2
=+ /-----+
| /-----+ \------+ 4
| | |
| | \------------+ 6
\------+
| /------------+ 3
\-----+
\------------+ 5
The tip numbers in the trees are corresponding with the columns of merged.alignment.csv, so 1 = Hordeum_vulgare,
2-3 = Triticum_turgidum 1-2, 4-6 = Triticum_aestivum 1-3. We can find that all the topologies are identical (i.e. ((A, D), B)) in fact, thus
we manually adjust the assignments according to the phylogenetic positions, by assigning chromosomes that is sister to Triticum_aestivum 3 as 1
and assigning the sisters of 1+3 as 2:
$ cat Triticum_aestivum.ancestor.txt
1A 1 4359 RoyalBlue 1
1B 1 4736 RoyalBlue 2
1D 1 4487 RoyalBlue 3
2A 1 5840 red 1
2B 1 6152 red 2
2D 1 5885 red 3
3A 1 5237 #99CC00 1
3B 1 5941 #99CC00 2
3D 1 5306 #99CC00 3
4A 1 3027 deepskyblue 1
4A 3028 3593 #339966 1
4A 3594 3907 fuchsia 2
4A 3908 4056 deepskyblue 1
4A 4057 4870 fuchsia 2
4B 1 3878 deepskyblue 2
4D 1 3582 deepskyblue 3
5A 1 4772 #339966 1
5A 4773 5450 deepskyblue 1
5B 1 5574 #339966 2
5D 1 5574 #339966 3
6A 1 4141 #FFCC00 1
6B 1 4627 #FFCC00 2
6D 1 4012 #FFCC00 3
7A 1 5573 fuchsia 1
7B 1 4892 fuchsia 2
7D 1 5419 fuchsia 3
$ cat Triticum_turgidum.ancestor.txt
1A 1 3906 RoyalBlue 1
1B 1 4136 RoyalBlue 2
2A 1 5192 red 1
2B 1 5463 red 2
3A 1 4956 #99CC00 1
3B 1 5832 #99CC00 2
4A 1 2869 deepskyblue 1
4A 2870 3347 #339966 1
4A 3348 3447 deepskyblue 1
4A 3448 4385 fuchsia 2
4B 1 3487 deepskyblue 2
5A 1 4057 #339966 1
5A 4058 4677 deepskyblue 1
5B 1 5037 #339966 2
6A 1 3651 #FFCC00 1
6B 1 4017 #FFCC00 2
7A 1 4880 fuchsia 1
7B 1 4244 fuchsia 2
We re-run the above commands (wgdi -pc, -a, -at):
 |
 |
|---|
Fig. 4. Refined subgenome assignments based on phylogeny evidence.
Now, all the topologies are identical:
chr1-3H:
/-------------------------+ 1 /-------------------------+ 1 /-------------------------+ 1
| | |
| /------+ 2 | /------+ 2 | /------+ 2
=+ /-----+ =+ /-----+ =+ /-----+
| /-----+ \------+ 4 | /-----+ \------+ 4 | /-----+ \------+ 4
| | | | | | | | |
| | \------------+ 6 | | \------------+ 6 | | \------------+ 6
\------+ \------+ \------+
| /------------+ 3 | /------------+ 3 | /------------+ 3
\-----+ \-----+ \-----+
\------------+ 5 \------------+ 5 \------------+ 5
chr4-6H:
/-------------------------+ 1 /-------------------------+ 1 /-------------------------+ 1
| | |
| /------+ 2 | /------+ 2 | /------+ 2
=+ /-----+ =+ /-----+ =+ /-----+
| /-----+ \------+ 4 | /-----+ \------+ 4 | /-----+ \------+ 4
| | | | | | | | |
| | \------------+ 6 | | \------------+ 6 | | \------------+ 6
\------+ \------+ \------+
| /------------+ 3 | /------------+ 3 | /------------+ 3
\-----+ \-----+ \-----+
\------------+ 5 \------------+ 5 \------------+ 5
chr7H:
/-------------------------+ 1
|
| /------+ 2
=+ /-----+
| /-----+ \------+ 4
| | |
| | \------------+ 6
\------+
| /------------+ 3
\-----+
\------------+ 5
The above processes (wgdi -pc, -a, -at) may be iterated more than twice to generate such a consistant phylogeny.
We analyze the gene retain (wgdi -r) patterns:
wgdi -r Triticum_aestivum-Hordeum_vulgare.conf
wgdi -r Triticum_turgidum-Hordeum_vulgare.conf
 |
 |
|---|
Fig. 5. Gene retain of subgenomes.
However, biased fractionation patterns to distinguish subgenomes are not observed.
Then we can build a final subgenome phylogeny:
cat Hordeum_vulgare.*.trees.nwk > Hordeum_vulgare.trees.nwk
astral-pro -i Hordeum_vulgare.trees.nwk -u 2 -t 8 -o Hordeum_vulgare.trees.nwk.astral
phytop -pie -cp Hordeum_vulgare.trees.nwk.astral
nw_display Hordeum_vulgare.trees.nwk.astral
Fig. 6. Subgenome phylogeny from phased results of WGDI. 1 = Hordeum_vulgare,
2 = Triticum_turgidum A, 3 = Triticum_turgidum B, 4 = Triticum_aestivum A,
5 = Triticum_aestivum B, 6 = Triticum_aestivum D.
- Genomic data (genome sequences in fasta format) of the allopolyploid complex are required.
- Homoeologous relationships of chromosomes are required. These can be obtained from above synteny analyses or whole genome alignments..
Here, we just use the example data [T. aestivum (AABBDD) and T. turgidum (AABB)] prepared in this repo:
git clone https://github.com/zhangrengang/subgenome_phasing_example
cd subgenome_phasing_example
cd subphaser
# download genome sequences of Triticum_aestivum
wget https://urgi.versailles.inra.fr/download/iwgsc/IWGSC_RefSeq_Assemblies/v2.1/iwgsc_refseqv2.1_assembly.fa.zip -c && \
unzip iwgsc_refseqv2.1_assembly.fa.zip && \
mv iwgsc_refseqv2.1_assembly.fa Triticum_aestivum-genome.fasta && \
gzip Triticum_aestivum-genome.fasta -f && \
rm iwgsc_refseqv2.1_assembly.fa.zip
# download genome sequences of Triticum_turgidum
wget -c ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/900/231/445/GCA_900231445.1_Svevo.v1/GCA_900231445.1_Svevo.v1_genomic.fna.gz -O Triticum_turgidum-genome.fasta.gz
Now, all the required input data are present:
$ tree
├── Triticum_aestivum-genome.fasta.gz
├── Triticum_aestivum-sg.config
├── Triticum_turgidum-genome.fasta.gz
└── Triticum_turgidum-sg.config
subphaser -i Triticum_aestivum-genome.fasta.gz -c Triticum_aestivum-sg.config -pre Triticum_aestivum_
subphaser -i Triticum_turgidum-genome.fasta.gz -c Triticum_turgidum-sg.config -pre Triticum_turgidum_
Then we need to check whether the genomes have been well phased and whether the identified potential exchanges are confident:
On the clustering heatmap (Fig. 6B) and PCA plot (Fig. 6C), a subgenome is defined as well-phased if it has clearly distinguishable patterns of both differential k-mers and homeologous chromosomes, indicating that each subgenome shares subgenome-specific features as expected.
SubPhaser outputs all windows whose enrichments do not match the subgenome assignments of their chromosome as potential exchanges, but further manual checks are still needed to identify them as bona fide exchanges. For example, at the 3’ end of wheat chr4A (Fig. 6D), the significant enrichments of subgenome B-specific k-mers are continuous (2nd from outer to inner circles), and the subgenome B-specific k-mers are as abundant as those on the chromosomes of subgenome B (5th circle) which contrasts to other subgenomes (4th and 6th circles). Combining all this information, we can confidently conclude that there has been an exchange, since the possibility of assembly errors has been ruled out with Hi-C data previously, etc. Since the distributions of subgenome-specific k-mers are usually uneven across the genome, inference should be careful and cautious to avoid type II errors.
 |
 |
|---|
Fig. 6. Subgenome assignments based on subgenome-specific kmers.
Here for comparison purpose, we convert the output of SubPhaser to the format of WGDI, to build the subgenome phylogeny using the same method (wgdi + astral-pro).
Link files for WGDI:
ln ../wgdi/*gff ../wgdi/*lens ../wgdi/*Hordeum_vulgare.conf ../wgdi/ak.txt ../wgdi/*Hordeum_vulgare.blockinfo.new.csv ../wgdi/pep.faa .
Convert the output of SubPhaser, with discarding segments < 5 Mb:
cat Triticum_aestivum_phase-results/Triticum_aestivum_k15_q200_f2.bin.group | grep -v "#" | awk '$6>=5{print $1"\t"$2"\t"$3"\t"$4}' > Triticum_aestivum.sg.bed
python ../script/subphaser2wgdi.py Triticum_aestivum.sg.bed Triticum_aestivum.gff Triticum_aestivum.ancestor.txt
cat Triticum_turgidum_phase-results/Triticum_turgidum_k15_q200_f2.bin.group | grep -v "#" | awk '$6>=5{print $1"\t"$2"\t"$3"\t"$4}' > Triticum_turgidum.sg.bed
python ../script/subphaser2wgdi.py Triticum_turgidum.sg.bed Triticum_turgidum.gff Triticum_turgidum.ancestor.txt
Build the subgenome phylogeny using WGDI:
wgdi -pc Triticum_turgidum-Hordeum_vulgare.conf
wgdi -a Triticum_turgidum-Hordeum_vulgare.conf
wgdi -pc Triticum_aestivum-Hordeum_vulgare.conf
wgdi -a Triticum_aestivum-Hordeum_vulgare.conf
paste Triticum_turgidum-Hordeum_vulgare.alignment.csv Triticum_aestivum-Hordeum_vulgare.alignment.csv | perl -pe 's/\t[^,]+//g' > merged.alignment.csv
echo "[alignmenttrees]
alignment = merged.alignment.csv
gff = Hordeum_vulgare.gff
lens = Hordeum_vulgare.lens
dir = Hordeum_vulgare.tree
sequence_file = pep.faa
trees_file = Hordeum_vulgare.trees.nwk
align_software = mafft
tree_software = iqtree
model = MFP
trimming = trimal
minimum = 4
delete_detail = true" > Hordeum_vulgare.conf
wgdi -at Hordeum_vulgare.conf
astral-pro -i Hordeum_vulgare.trees.nwk -u 2 -t 8 -o Hordeum_vulgare.trees.nwk.astral
phytop -pie -cp Hordeum_vulgare.trees.nwk.astral
nw_display Hordeum_vulgare.trees.nwk.astral
Fig. 6. Subgenome phylogeny from phased results of SubPhaser. 1 = Hordeum_vulgare,
2 = Triticum_turgidum A, 3 = Triticum_turgidum B, 4 = Triticum_aestivum A,
5 = Triticum_aestivum B, 6 = Triticum_aestivum D.
If you use this pipeline, please cite:
Zhang RG, Shang HY, Jia KH, Ma YP. Subgenome phasing for complex allopolyploidy: case-based benchmarking and recommendations [J]. Brief. Bioinform., 2024, 25 (1): bbad513 DOI:10.1093/bib/bbad513