Data dimensions after fold()

7 views
Skip to first unread message

Maycon Oliveira

unread,
Jan 19, 2026, 1:29:36 PM (2 days ago) Jan 19
to dadi-user
Hello all! I was wondering, what happens to the sample sizes after i call fs.fold() within dadi? To my knowledge, the sample sizes should be updated to reflect the dimensions of the new SFS, but after doing some tests I realize this is not the case currently. Do I have to manually project down the SFS after folding in order to account for this?

Ryan Gutenkunst

unread,
Jan 19, 2026, 3:37:42 PM (2 days ago) Jan 19
to dadi-user
Hello Maycon,

Folding does not change the sample size, because it is still the same samples. dadi tracks whether a spectrum is folded or not internally and automatically adjusts the model to reflect that. You should not need to do anything to adjust for folded versus unfolded data.

Best,
Ryan

On Jan 19, 2026, at 9:20 AM, Maycon Oliveira <oliveira...@gmail.com> wrote:

Hello all! I was wondering, what happens to the sample sizes after i call fs.fold() within dadi? To my knowledge, the sample sizes should be updated to reflect the dimensions of the new SFS, but after doing some tests I realize this is not the case currently. Do I have to manually project down the SFS after folding in order to account for this?

--
You received this message because you are subscribed to the Google Groups "dadi-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dadi-user+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/dadi-user/156a8831-4fc9-459b-a764-eff768fb2644n%40googlegroups.com.

Maycon Oliveira

unread,
Jan 20, 2026, 6:31:05 AM (yesterday) Jan 20
to dadi-user
Oh, ok. I got confused because on one of my results I got the following (attached) plots which look very sparse, but I guess that's due to many bins with missing counts on my SFS, right? In that case, I believe the correct approach would be to project down the SFS, is that a correct assumption?
cc_ei_sec_contact_sym_mig.pdf

Ryan Gutenkunst

unread,
Jan 20, 2026, 6:24:02 PM (yesterday) Jan 20
to dadi-user
Hello Maycon,

Yes, the sparseness is due to many zeros in your data. You can project down to fill it out for plotting purposes. You don’t need to project down for the analysis itself.

Best,
Ryan

Reply all
Reply to author
Forward
0 new messages