Question about the threshold in Manhattan plot

[foala]

Hi, Thanks for this great tool. So, i generated a Manhattan plot for my population using 99.9% as a threshold.

Can you please assist in interpreting the threshold? Eventhough it is 99.9%, it still includes all of the snp windows. Is it correct to detect outliers?

Thank you again,

[alachins]

We typically use the top 0.05%, so 99.95% as input threshold to RAiSD.

[charlescongxu]

Hi @alachins Is there a reason why the top 0.05% is considered significant? I am trying to use RAiSD on metagenomics data and some genomes are poorly covered so I have few sporadic SNPs. In these cases, setting a 0.05% heuristic seems inappropriate?

[alachins]

Hi Charles, Based on the small number of SNPs that you have, I would not recommend using RAiSD for the analysis because all methods implemented in RAiSD are SNP-driven. You could try SweeD and/or OmegaPlus instead. Best regards, Nikos A.

On Tue, Jan 25, 2022 at 3:29 PM Charles Xu @.***> wrote:

Hi @alachins https://github.com/alachins Is there a reason why the top 0.05% is considered significant? I am trying to use RAiSD on metagenomics data and some genomes are poorly covered so I have few sporadic SNPs. In these cases, setting a 0.05% heuristic seems inappropriate?

[image: image] https://user-images.githubusercontent.com/17847284/150995565-aabf7e29-9793-41c2-9f54-85504fd3550d.png

— Reply to this email directly, view it on GitHub https://github.com/alachins/raisd/issues/21#issuecomment-1021238745, or unsubscribe https://github.com/notifications/unsubscribe-auth/AALKWCUY7VQX24ASYPMPAYLUX2XTVANCNFSM4OFX4AYA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you were mentioned.Message ID: @.***>

-- Nikolaos Alachiotis

[charlescongxu]

Hi Nikos,

I see. I do have some metagenomes that have lots of SNPs. In these cases, how should the threshold be determined?

[alachins]

Hi Charles, There is no reason behind choosing 0.05 to be considered significant. It is just something usually done. Normally, you would have to find the demographic model, perform neutral simulations, obtain a threshold value (controlling for the FPR), and use that value as threshold for your data. The easy workaround commonly used is to directly take the top x% of your scores as the threshold. The rationale here is that the majority of your genome is neutral since sweeps act in a narrow, localized manner and do not affect the whole genome. Best regards, Nikos A.

On Wed, Jan 26, 2022 at 3:31 PM Charles Xu @.***> wrote:

Hi Nikos,

I see. I do have some metagenomes that have lots of SNPs. In these cases, how should the threshold be determined?

[image: image] https://user-images.githubusercontent.com/17847284/151181922-92d1f1e8-94d4-48ff-90b0-0ab1e9940a8f.png

— Reply to this email directly, view it on GitHub https://github.com/alachins/raisd/issues/21#issuecomment-1022252241, or unsubscribe https://github.com/notifications/unsubscribe-auth/AALKWCSXBUWK2PMSNW2AZCLUYAAU7ANCNFSM4OFX4AYA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you were mentioned.Message ID: @.***>

-- Nikolaos Alachiotis