Let me pin the link tot the list of tutors so ill find it in a few weeks: https://docs.google.com/spreadsheets/d/1W2nUsdp64zugXmUZ0roVazjeJ39OyASt0OvTpjBP2Dg/edit?usp=gmail#gid=0
Once you enter your room today, please fill out the following survey. https://forms.gle/xUd8ZUocVxrF2DzTA
Who wants to have point-click interface for downloading files, use FileZilla for PC/Linux. Set it up as SFTP with SSH credentials. Then login, navigate one folder up to "home" and there you see the faculty members' folders
Hi Session B! Once you enter your room, fill out this survey https://forms.gle/5owbJcn5gDbvM7yc6
Hi all, I'm just going to use this channel to shamelessly (ok with a little shame) plug this recent review of ours on the history, current state, and future of GWASs of human behaviour. We wrote it to be accessible to a wide audience, such as the group of students of this course. Hope you enjoy it!
*Thread Reply:* It’s a great question and was also asked anonymously — I took a stab at it here!
Hi all - we are hiring a research scientist in statistical genetics and psychiatric epidemiology. Further details in the ad attached here. Please let me know if you have any questions or would like to learn more!
Here is a short video to show how to copy the files for day 5 (thanks Michel Nivard for the suggestion)
Good morning/evening/night later today (in 2.5 hrs) we are in the <#C020248GYLT|day06-genomicsem> channel, @Danielle will be in charge of assigning breakout rooms, directing tutors to students in need etc, I will present some slides (@Andrew Grotzinger will do so in the B session). Its going to be fun, we don't expect you to get it all done. we actually added assignments so you can get some further practice in after the workshop ends if you want to. I share this video which walks trough getting all the data for the session, your tutors for the A session are:
1 @Perline Demange 2 @Javier de la Fuente 3 @Jackson Thorp 4 @Andrea Allegrini 5 @Margherita Malanchini 6 @Mark Adams 7 @Gunn-Helen Moen
}
FYI: maybe you all already know (I saw some tutorials doing it), and I am stating the obvious, but for those who don’t, a nice hack in Rstudio to make life easier is to create named code subsections. You can do it just by following a commented line with four signs (i.e., ####). For me, it was game-changing. 😅
Might any twin modeling folks in the channel have a "twin family model with specified shared environment" script (or anything close to this that I could work from)? I am interested in doing something like this to understand the (measured) environmental mechanisms through which genetic risk is transmitted in families.
For tomorrow, head over to this channel for Sequencing and Hail! 🙂 We posted an announcement to prep you for the excitement that is tomorrow’s workshop session @channel
}
@channel Behavior Genetics 2021 Annual Meeting Focus on applied methodology in Behavioral and Statistical Genetics Online 24 hour zoom meeting. Starts 7pm 28th June US-Eastern = midnight 28/29th June UK = 9am Aus-Eastern Live talks + interactive poster sessions. Join for the sections you are interested in (watch recorded sessions after) Registration by donation (suggested $24) For details see https://bga.org/current-meeting-2021/
*Thread Reply:* there is some discussion here: https://boulder-workshop.slack.com/archives/C01TLU7JAQK/p1623763267118400
}
*Thread Reply:* this isn’t naive and is an excellent question. For common variants, given the level of LD we observe in modern human populations, there are roughly 1 M independent SNPs across the genome and thus dividing .05 by 1M = 5e-8 is a justifiable alpha level. Of course, this is a rough approximation and differs a bit between populations (e.g., Africans have lower LD and thus need a slightly more conservative correction than Europeans, but this isn’t a huge difference and these thresholds, such as .05, are ultimately arbitrary anyway). When it comes to rare variants, the story is quite different. Rare variants tend to have much lower levels of LD, even with other rare variants, than do common variants with other common variants. Thus, as the MAF gets lower (e.g., if you’re using a v. large sequencing ref panel to impute your data down to, say, MAF = .0001, which is quite possible now), your GWAS will have more than 1M independent tests and you therefore need to use a more conservative correction. I’ll post a paper about this below. An attendant problem with a more conservative threshold for rare variants is lower power. As variants you want to test become increasingly rare, it may make sense to reduce teh number of tests performed by using gene-based or pathway-based association tests, or by using alternative approaches such as a Bayesian one that allows for weighting based on annotation.
Wu, Y., Zheng, Z., Visscher, P.M. et al. Quantifying the mapping precision of genome-wide association studies using whole-genome sequencing data. Genome Biol 18, 86 (2017). https://doi.org/10.1186/s13059-017-1216-0
Fadista J, Manning AK, Florez JC, Groop L. The (in)famous GWAS P-value threshold revisited and updated for low-frequency variants. Eur J Hum Genet. 2016;24:1202–5.
*Thread Reply:* Great discussion here. I agree with Matt. Not a naive question at all. Just want to add this paper too as relevant. Note that this is based on sequenced data and not imputed SNPs as we normally do, which has an impact on false positive rates as shown in Wu et al. Still a relevant paper in particular to illustrate differences between ancestries. https://www.nature.com/articles/jhg201672
