Search results

kinda illegal

damn 33 bucks for that? i would do it for half...

Breton is extremely common for everyone now, I got 3.3% eventho im from Spain

Posting just before the supposed update, as of now the results dont make much sense

Im using eigenstrat format files and I tried not using low snp count samples

I'm getting horrible results when making PCAs with ancient samples, Im using the v62 1240k dataset they all bunch up in the same place here i changed the numoutevec number to 15 and added modern samples(red square) too again the same problem, they all plot more or less together not making...

any text editor, pic related is Kate which is the default one for KDE

That error you get is a Linux basic, its because the spaces, instead of "John Smith" try "John_Smith" or "JohnSmith" and just modify the outer column, the "control" one

it doesnt have to be any particular version, im using the latest one (4.4.2) and it works just fine

uh, now I understand something... yeah thats what they do, here you have a lot of modern (and ancient) samples from other studies https://reich.hms.harvard.edu/datasets check supplementary data from studies, for example (from...

@Jovialis did you delete a post regarding numoutevec setting having to be 15? what does it exactly do?

kinda unrelated but what do u think about v54.1 1240k vs v62 1240k? v62 is ofc more up to date but I see weird samples and labels and the v54 seems more curated or "refined"? which one do you use? I get slight differents results

hey, got it working first try but it didnt add me edit:

At the end I did the merge with both plink and eigensoft, the two work okay. Where did you get hold of v62.0? edit: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/FFIDCW

haha you are a bit confused, the Z-score in qpAdm is calculated by dividing the weight between the standard error(calculate it yourself, you will see how it matches), so of course you want the z score to be high! imagine your model gets 0.54 or 54% of Turkey_Barcin and only 0.0180/1.8% of SE...

Where did you read that? Z-score should be over 1.5 or 2 to be significant, the more the better, more Z-score = more robustness/certainty. It estimates how much given ancestry proportions differ from the expected value (https://en.wikipedia.org/wiki/Expected_value). I think its impossible to get...

its a p value extremely close to 1 good? seems a bit fishy to me, I got high standard errors tho so maybe thats why

hmm, do you have the same number of SNPs in your merged files of different formats? my merged v54.1.p1_HO.bim file got 597k SNPs, pretty good considering my raw DNA data from MyHeritage had a bit over 600k SNPs, but the .snp file that I got as a result of merging with v54.1.p1_1240K and...

I have so many questions... but I already spammed this thread enough