• Add a column to output that includes observed frequenceis

Hello,

I am trying to apply SoFoS on triallelic data and calculate folded SFS on a designated allele. I used the v2.0 on such data and encounter an error:

#SoFoS v2.0
#date=2021-12-15T17:41:07-0600
#epoch=1639611667475
#path=synLTR.triAllelic.vcf
#alpha=1
#beta=1
#size=10
#folded=0
#refalt=1
[E::bcf_calc_ac] Incorrect allele ("2") in Ki11 at 1:1

If all loci are triallelic homozygous, i.e., 0/0 (ref), 1/1, and 2/2, and I want to estimate folded SFS on allele 1, is there a way to do so? Is it equivalent to converting 2s to 0s, and estimating the folded SFS for 0s and 1s?

Thanks,
Shujun

The assumption is that non-variant sites are included in the input. Does a beta prior still fit a situation when the user only provides sites with polymorphic data?

Hi,
I was wondering if SoFos can produce multidimensional SFS and what data could it require. Can you give me some suggestions?

Very many thanks.

Hello,

Thank you for developing SoFoS which is very helpful to control population size especially for small samples!

I just want to make a note here that the -n parameter also means the resampled population size (especially true for diploids). For example, if pop 1 has 37 samples and pop 2 has 100 samples, setting n = 30 controls both populations to have a resampled size of 30. It took me some readings including all posted issues, a manuscript draft, and some twitter discussions to come to realize the meaning of -n. Hopefully, this note will help others.

Best,
Shujun

• Add .circleci/config.yml to repo.

• Support reading GP tags.
• Support reading AC/AN tags.

Hello @jporton,

I used -r to specify using the REF as the ancestral state, and found SoFoS was not taking this parameter for some occasions. This seems to be related to the INFO column of the vcf file where there is AC (allele count) information. When removing it or replacing the AC information, it takes the -r parameter again.

This test.vcf.zip can reproduce the issue:

sofos -n 10 -u -r -a 1.0 -b 1.0 test.vcf 
#SoFoS v2.0
#date=2021-12-18T16:33:38-0600
#epoch=1639866818307
#path=test.vcf
#alpha=1
#beta=1
#size=10
#folded=0
#refalt=1
#
## [0.0006s elapsed] 60 sites processed
Number,Prior,Observed,Posterior
0,5.4545454545454959,0,0.69853641197197858
1,5.4545454545454959,1,1.3048755954406486
2,5.4545454545454959,1,1.8488491162524154
3,5.4545454545454959,1,2.4702060322842092
4,5.4545454545454959,0,3.2635682519285871
5,5.4545454545454959,1,4.2688676722197485
6,5.4545454545454959,0,5.4985717806044025
7,5.4545454545454959,0,6.9696393737318694
8,5.4545454545454959,0,8.7375634335076597
9,5.4545454545454959,0,10.95262220018968
10,5.4545454545454959,56,13.986700131869155

sed 's/AC=/AA=/' test.vcf | sofos -n 10 -u -r -a 1.0 -b 1.0 - 
#SoFoS v2.0
#date=2021-12-18T16:34:02-0600
#epoch=1639866842418
#path=-
#alpha=1
#beta=1
#size=10
#folded=0
#refalt=1
#
[W::vcf_parse] INFO 'AA' is not defined in the header, assuming Type=String
## [0.0006s elapsed] 60 sites processed
Number,Prior,Observed,Posterior
0,5.4545454545455394,0,16.557384357674806
1,5.4545454545455394,55,18.025665060544522
2,5.4545454545455394,2,12.323574756722911
3,5.4545454545455394,1,6.569310900606677
4,5.4545454545455394,0,2.9275361416968226
5,5.4545454545455394,0,1.1301463637682869
6,5.4545454545455394,0,0.41433400150816324
7,5.4545454545455394,0,0.22850857449687428
8,5.4545454545455394,0,0.30356657899338751
9,5.4545454545455394,1,0.54636205470466404
10,5.4545454545455394,1,0.97361120928371347

Any ideas? Thanks!

Shujun

• Use the AA tag instead of REF to identify ancestral alleles
• Default to reporting folded results
• Add a flag to use REF instead of AA when doing unfolded analyses.

Hi,
I download SoFos from Github, but I don't know how to install, can you give me some suggestions?

Thanks.

Should sites with 100% missing data be skipped or included in the posterior based only on prior?