Random introduction of "-"

[Viloleal]

Dear everyone,

I have been using Rhierbaps for some months now and have recently realised a weird behaviour that I would like to point out.

I have loaded a multialignment fasta file using the load.fasta function and when I checked the dataframe I realised that there were "-" along the different samples. This at first wasn't strange, since there are some samples that have one or two ambiguous calls like K or R. However, what called my attention was the fact that the "-" were appearing in more samples, including those that did not have any ambiguous calls. In the example below, you can see that sample 10 has a "-" at position n3, whereas in the msa this should be a "T".

I asked a colleague of mine to repeat this using her computer and she got the same results. She then checked the dataframes from her data and realised that this happens as well on her data:

Again, there is a "T" at position n1 that is being called as a "-". We have also seen that this happens as well with "Gs".

This appears to happen randomly throughout the data frame, but the errors do appear in the same places each time we run the load.fasta function. We would like to know if you have seen this behaviour before and know what could be the cause?

Our Rhierbaps version is 1.1.4 and R version are 4.2.3 and 4.1.3, respectively.

Thanks!

[gtonkinhill]

Hi,

Thanks for reporting this issue. To resolve it faster, it would help a lot if you could provide a small reproducible example including the fasta file and the exact commands you ran that led to the problem.

[Viloleal]

Hi,

Here I send you a test file containing two of the sample alignments and the root. The commands I ran were:

library(rhierbaps)
library(ggtree)
library(phytools)
library(ape)
set.seed(1234)

setwd(dir = "C:/Users/myworkingdirectory")
fasta.file.name <- "test_file.fasta"
snp.matrix <- load_fasta(fasta.file.name)

Now there are many more "-".

I must say that I just tried running:
snp.matrix <- load_fasta(fasta.file.name, keep.singletons = TRUE) and the "-" have disappeared, both in the test and the original file:

Thanks

EDIT: Sorry, I forgot to upload the fasta file.
test_file.zip

[gtonkinhill]

Hi,

Thanks very much for providing an example. As you've pointed out this is caused by the keep.singletons = FALSE setting.

Admittedly it is not very intuitive, but rhierbaps converts SNPs observed only once into '-' characters when this option is set. These columns of the alignment are then filtered out by the algorithm when the hierBAPS function is called.

We recommend filtering out these columns as they are not very informative of population structure.

I would also recommend taking a look at fastbaps which is much faster and often leads to better results.

[Viloleal]

Hi,

Thanks for checking it, i'm glad to hear it's just normal functioning of rhierbaps. I agree that it is not very intuitive, since I always thought that rhierbaps removed singletons from the start, not masked them. It makes sense that it removes the columns afterwards.

Thanks again for your help!

Kind regards