question about indels

[rderelle]

Hi,

Thanks for developing phastSim - a lovely tool.

I just have a quick question regarding how phastSim adjusts genomic positions when insertions and deletions are added to the genomes: when subsequent mutations occur, do their genomic positions correspond to the 'new positions' (ie, adjusted with old indels), or those of the original genomes. If the latter, it means that phastSim cannot mutate old inserts and do make 'mistakes' by mutating positions that do not exist due to old deletions.

That is not a major problem in both cases, but I would need this information to correctly reconstruct SNP alignments.

Many thanks if you could help me.
Romain

[NicolaDM]

Hi Romain,

I don't remember the exact indexation we used, but the idea is that previous insertions can still be deleted and have new additional insertions within them, as you would expect. In short, it's the usual indel model you would expect in a non-approximate simulator (as in INDELible) and I think the indel model is called called GGI model (see https://academic.oup.com/mbe/article/41/9/msae177/7739074 ).
You can find more information in the phastsim paper: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1010056

[rderelle]

Hi Nicola,

Many thanks for your reply and these references. From the phastSim paper, it seems that the genome tree is modified every time an indel occurs, meaning that phastSim cannot mutate a deleted position but could mutate an inserted position, which is great. However the indexing does not seem to be modified ("Node.position" appears in algorithm 2 but not in algorithms 5-7 mentioning indels). Thanks, and again, not a issue per se!

Romain

[willboulton]

Hi Romain - just to let you know, on the point of the indel indexing, it starts becoming a little complicated but basically as soon as indels are generated, they have their own set of genome coordinates, so the positions on the base genome don't get modified. Instead, each new indel has its own set of coordinates.

For example if the genome is ACGTA and then position 2 is substituted C->G, and then an indel AAA is generated at position 3 (so now the genome is AGGAAATA), and then there's a substitution at position 4, this refers to T in AGGTA. If there was a new substition on the indel (say at the third coordinate), this would be referred to as at position 1,3 (coordinate 3 of the 1st indel).

For deletions, you can just think of deletions as substitutions to the symbol "-" that can't substitute back.

Not sure if this clarifies things, as the whole procedure is a bit complicated, but hope it helps - if not then ignore this message :)

[rderelle]

Thanks a lot Will, and my apologies for this late reply. Your example did indeed clarify things.
I believe my script now correctly reconstruct SNP alignments based on these info.

Thanks both for your help.