I am working on intermediate wheatgrass, which is allohexaploid. But its sub-genome donors are still unknown. Many attempts have been made to identify the donors, and the results have been quite mixed. Now we can discover genome wide markers using GBS. These markers can be used for phylogeny analysis. The phylogenetic inference might help to determine the sub-genome donors.
Anyone has comments about it? Is it feasible?
Dear Dr. Zhang,
I think that it is feasible (we are doing something similar) keeping in mind important methodological considerations:
1) GBS is a coverage dependent technique. It is needed enough coverage (at least 5X/diploid genome, more is better) to call SNPs in a reliable way. For an hexaploid genome, it will need 3 times more coverage than in a diploid genome, to be able to separate the homoeologous reads and still be able to have enough coverage.
2) Also the pipeline is more reliable and easier if you have a reference. If you don't you can use UNEAK (http://www.maizegenetics.net/images/stories/bioinformatics/TASSEL/uneak_pipeline_documentation.pdf) but it is more difficult to identify the subgenomes.
3) If you want to do a phylogenetic analysis, you should be able to identify the homoeologous reads/SNPs and assign them to a subgenome, but it is difficult if you don't know the polyploid progenitors. In our case we used a population genetic approach to a) check if a polyploid follows an admixture model, b) check the progenitor origin compared with diploid populations. Once we have identified the most probable origin, we reconstructed a "consensus reference diploid progenitors" and we identified/separated the polyploid subgenomes based in a preferential mapping.
Please let me know if you want to have more details about the process,
Sincerely,
Aureliano
Dr. Bombarely,
Thank you very much. Our data also showed that UNEAK works well in SNP calling for hexaploid wheatgrass. But for our full-sib F1 poplation, we have to discard tons of SNP markers due to their improper segregation (mainly stem from homoeologous loci).
I have been thinking about dissecting the sub-genomes of wheatgrass and determining the donors, which might help to assign wheatgrass SNPs to subgenomes. Your third point is valuable for me. Could you please give me some references or more details about the analysis.
Best,
Xiaofei
Dear Dr. Zhang,
In our cases we used Structure (http://pritchardlab.stanford.edu/home.html) and FineStructure (http://paintmychromosomes.com/) to analyze the ancestral population structure of the diploid/polyploid complex. To do that you need:
1) to transform the HapMap output format of UNEAK into the Structure or Phase format used by these programs respectively. I have written two scripts to do it (https://github.com/aubombarely/GenoToolBox/tree/master/PopGenetics).
2) Run these pipelines. FineStructure is specially interesting because you can visualize the results as a Heatmap or a PCA.
The second part can be more complicated if you don't have a reference because probably you'll need to write an script to assign reads based in similarities with other reads that you define as "progenitors". If you have a refence you can:
1) Map diploid progenitor GBS reads with your reference using Bowtie, BWA or Novoalign.
2) Use samtools and pileup to generate a consensus sequence (http://samtools.sourceforge.net/samtools.shtml#7), one per progenitor.
3) Join all the progenitor consensus sequences.
4) Map your polyploid reference reads with the progenitor's reference. Reads should preferentially map to its closer diploid progenitor. For example if you have two progenitors A and B, and you map your polyploid P, a read can a) not map, b) map equally to two positions A1 and B1, in which case they will have the same mapping score (AS, XS), or c) map preferentially to one subgenome in which case AS > XS. Based in this you can use the sam file and this script (https://github.com/aubombarely/GenoToolBox/blob/master/SeqTools/SeparateHomeolog2Sam) to separate the reads based in the preferential mapping of the genomes. It is recommended to filter the low mapping score reads. Once the reads have been asigned to each of the subgenomes, you can get specific SNPs and do any no-gene-based phylogenetic analysis to decipher a more clear relations between diploids and polyploids.
I hope this helps, please let me know if you need more information.
Best
Aureliano.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools