poruloh / Eagle

Haplotype phasing software
63 stars 15 forks source link

Getting phase confidence for each SNP? #40

Open EmiliaCXY opened 1 year ago

EmiliaCXY commented 1 year ago

Hi Po-Ru!

I was wondering if it'd be possible to get a phasing confidence value for each SNP? I was trying to calculate a haplotype frequency based on a set of phased SNPs and thought it would be beneficial to incorporate the probability of phasing into my calculation. I saw in the output log, there's a mean phase confidence for each individual, so I was wondering if there's anywhere I could access per SNP confidence.

Thanks for your time and efforts!

Best, Emilia

poruloh commented 1 year ago

Hi Emilia,

Generally phasing confidence is difficult to assess at a per-genotype level and is also complicated to use (because phasing is a property of pairs of heterozygous variants rather than a single variant). For these reasons Eagle doesn't output genotype-level measures of phasing confidence, and the individual-level estimates that are reported are only rough estimates. However, phase quality is generally uniformly high across the genome, so it ought to be ok not to bother filtering on phase confidence.

Best, Po-Ru

On Sun, Jul 23, 2023 at 10:20 AM EmiliaCXY @.***> wrote:

Hi Po-Ru!

I was wondering if it'd be possible to get a phasing confidence value for each SNP? I was trying to calculate a haplotype frequency based on a set of phased SNPs and thought it would be beneficial to incorporate the probability of phasing into my calculation. I saw in the output log, there's a mean phase confidence for each individual, so I was wondering if there's anywhere I could access per SNP confidence.

Thanks for your time and efforts!

Best, Emilia

— Reply to this email directly, view it on GitHub https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_poruloh_Eagle_issues_40&d=DwMCaQ&c=WO-RGvefibhHBZq3fL85hQ&r=pBlP9cuNT3BUDsRDLkVbgWxzX8j_R7x0nL_Q4UWFMPo&m=HGNU7_-Y1y1jEMMbHxJbXu5b8Nw2UCDyeWbYUkcS0Fhm32GRBLfaRtLQDsZlvPsh&s=QZkh0ONdWhKqlOxLIBeuYvlGMDkXlWqtU-VWMrySHiw&e=, or unsubscribe https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_ACDGGQLMBQBSOYMIE5XRYKDXRUXKJANCNFSM6AAAAAA2URQK5E&d=DwMCaQ&c=WO-RGvefibhHBZq3fL85hQ&r=pBlP9cuNT3BUDsRDLkVbgWxzX8j_R7x0nL_Q4UWFMPo&m=HGNU7_-Y1y1jEMMbHxJbXu5b8Nw2UCDyeWbYUkcS0Fhm32GRBLfaRtLQDsZlvPsh&s=sjAiZc7HhmJpbwZR4gTUmU09xF9IMnxviqin-rhaTKk&e= . You are receiving this because you are subscribed to this thread.Message ID: @.***>