Hi all,
I'm trying to construct a phylogeny for a plant diversity panel from a set of WGS resequencing data (7-10x coverage). I have an initial VCF generated by mapping data from approx. 340 accessions from "species A" against a species A reference genome. The VCF also includes data from 10 "species B" accessions from the same genus. I'd like to use this group to root my phylogenetic tree later.
My question is how to treat these species B accessions during filtering. They, predictably, have a higher fraction of missing data (~20% variants) than the species A accessions (vast majority < 5%). Getting rid of any individuals with more than 10% missing data would be a conventional starting point but this doesn't seem appropriate or useful.
One suggestion I have seen was to conduct all filtering and linkage pruning purely based on species A statistics, then merge in species B calls just prior to tree-building, accepting their higher level of data missingness. This seems sensible, but I can't find much support for this approach, making me rather sceptical.
Any ideas / suggestions / literature would be really appreciated.
Many thanks, Max
•
written
by
maxrwjones •
0 answers
No answers yet.
Log in to answer this question.
-
Including reference genome in SNP-based phylogeny
written by maxrwjones •
-
Phylogeny from mixed GBS and WGS
written by maxrwjones •
-
Testing ghost species using f3 statistics
written by ki-woong.nam •
-
Help with PLINK
written by Dhara Awasthi •
-
PLINK input files from Genome Studio
written by matziast •
-
Theory question on dN/dS values for phylogeny
written by DNAngel
-
Calculate individual missingness in vcf NOT using VCFtools
written by evan.hersh •
-
Statistical significance in phylogenies
written by ceruleanivy •
-
Why Plink stoped analyzing without make a new PED file and MAP file?
written by Bahar •
-
How To Create Genome-Wide Estimation Of Ibd Sharing Using Plink From Tped Files?
written by Kevin
I don't have an answer to your question, but I'd like to hear your view on the waster tool - https://github.com/chaoszhang/ASTER/blob/master/tutorial/waster.md
I use it to generate SNP based trees from many genomes, but it can work on raw fastqs too. I wonder if it could be helpful on your data - it is certainly easy to run.
Hi Colin,
Thanks for the response. This looks like a really interesting tool that I had not come across. I have mostly used IQ-TREE2 in the past. Do you know if this is suitable for constructing a tree for a diversity panel (i.e. mostly within a species) as I am trying to do? (I guess they use the term "species tree" to distinguish whole-genome comparisons from per-locus "gene trees" and therefore comparing within a species is effectively no different to comparing between species?)
Also seems odd that it can't take paired FASTQ as input - it wants me to concatenate the mate pair 1 and mate pair 2 files together. This seems like it's throwing away information, but I guess if it is truly "without-alignment" then the positional information provided by mate pairs is not needed?
Many thanks, Max
It is a kmer approach, yes. You can do some fastq merging with bbmerge - see alternative AI docs here - https://deepwiki.com/chaoszhang/ASTER/3.5-waster
I have only used waster (from assembled genome seqs, not read fastqs so far at least) for intraspecies comparisons. I have noticed it is not robust in creating trees when comparing near identical (just a few insertions, few SNPs) genomes. Try it out for a few of your (subsampled) sequences and see if it looks decent before progressing further.
edit - Bootstrap values are generated too with this tool.