FAANG 2022 Workshop Zoom Chats

Poster/Lightening Talk session

02:56:38Huaijun Zhou:https://app.gather.town/app/FZGSmMglvw1zeSms/FAANG%202022
03:02:28Dan Freeck:I'm not sure how to see the posters. I'm stuck where the last presentqation ended. and see only Chris
03:03:19Emily Clark:Here’s the link again:
03:03:22Emily Clark:https://app.gather.town/app/FZGSmMglvw1zeSms/FAANG%202022
03:03:35Peter Harrison:Hit X to see the posters
03:43:09Andreas Pfenning:I’m having trouble hearing
03:43:10Doreen Becker:Audio is very bad
03:43:27Eric Veien:Can't hear her very well
03:43:38Eleonora Cappelletti:There is no audio
03:43:38Doreen Becker:Audio is bad
03:43:48Sylvain Foissac:Indeed
03:43:58Maria Francesca Piras:No audio…
03:44:17Eleonora Cappelletti:If you prefer I can present it now
03:45:38Andreas Pfenning:That’s better!
03:54:02Dominique ROCHA:@Alex: why you didn't try methyl-seq, the enzymatic methylation technique developed by NEB?
03:54:25Dominique ROCHA:@Alex: for WGBS, 19x coverage was before or after mapping?
03:57:39Dominique ROCHA:@Alex: I was talking methylation analysis using methylation sensitive enzymes. Methyl-seq is different. Thanks Alex
04:00:47Dominique ROCHA:@Alex: Oops I meant "I was NOT talking about methylation analysis using methylation sensitive restriction enzymes". Sorry for the typo. My fingers are too big for my keyboard. :((
04:01:03Ruidong Xiang:Hi, Huaijun, am I at the right channel?
04:03:05Emily Clark:@Ruidong I have transferred you to be a panelist.
04:04:20Alex Caulton:@Dominque, Sorry I just read that it was and "enzyme-based alternative to bisulfite conversion" so assumed it was a methylation sensitive enzyme approach, I have had a further look into it. It does look interesting, have you tried it with much luck? We are getting accurate results with the current RRBS methodology but the lab workflow is not particularly flexible as libraries can only be stored short term due to the bisulphite treatment.
04:05:54Dominique ROCHA:@Alex: thanks for your reply. No I didn't have a chance yet to try methyl-seq. :((
04:06:10Dominique ROCHA:Sorry I can't hear the recording!
04:06:28Christa Kühn:I agree, no audio
04:06:46Dominique ROCHA:Thanks I thought was my ear wax. :((
04:09:45Dominique ROCHA:@Loan: sorry it was difficult to hear your recording. I would assume that you could do the same work using the CpG mammalian array Alex talked about before, S. Horwath has used it in many mammalian species to develop an age clock. Have you tried this approach?
04:09:55Andreas Pfenning:@Loan, How do you determine accuracy? Is it correlation between predicted and actual age. Also, do you separate training and test groups?
04:10:04Frederique Pitel:@Dominique @Alex The platform in Toulouse tested the NEB Methyl-Seq, and we have great results, at least in quail :-)
04:11:22Dominique ROCHA:@Pitou: muchas gracias ;)
04:13:06Alex Caulton:@Dominque, we have used the methylation array to develop an epigenetic clock for ruminant species (sheep, goat, cattle and deer) it has high accuracy (r > 0.97) but as Loan pointed out the costs associated with this assay are higher than nanopore. Our clock paper can be found here: https://www.mdpi.com/2073-4425/13/1/96
04:14:56Dominique ROCHA:@Alex: thanks a lot (your paper will be my next bedtime reading)
04:17:43Dominique ROCHA:@Andreas: your method uses conservation. Did you check how many species-specific enhancers do you miss?
04:19:00Carrie Finno:@Andreas- fantastic talk! will the dataset of predicted functionally conserved elements be publicly available post-publication?​ It would be quite useful for our animal GWAS studies too!
04:19:39Andreas Pfenning:@Dominique: Good question. It depends on how close of a species we have for annotation. For example, when we had the macaque genome, we did not miss many human enhancers. However, adding species without close relatives in our dataset, like cow, improved things substantially.
04:19:41Loan Nguyen:@Andreas: We assessed the accuracy of predicting age using a cross validation strategy. Sets of 5 randomly animals had their age removed from the analysis, but they were still included in the methylation relationship matrix. This will resulted in age effects being predicted for these animals. These age effects were then correlated with the actual age for these animals. The cross-validation was repeated 10 times until all animals had been dropped from the analysis but included in the validation. the correlation of predicted age and actual age was taken as the accuracy of the prediction age.
04:20:32Elisabetta GIUFFRA:@All: ask your question in chat!
04:20:46Dominique ROCHA:@Andreas: thanks. What about sheep/goat? They're quite close.
04:21:06Dailu Guan:If it’s possible to train model using multi-chipseq data? Do you have any idea about that?
04:21:41Andreas Pfenning:@Carrie - Thanks! We’re working with UCSC to make them available. If you would like them, we can share them with you. (Email apfenning@cmu.edu)
04:22:38Andreas Pfenning:@Dominque: We make predictions in goat, but haven’t used it in training yet.
04:23:02George Liu:@Andreas: Great talk! We are interested in your methods. Will touch base with you soon.
04:23:47Andreas Pfenning:@Dailu: It depends on the specific ChIP-Seq. For broad peaks (like H3K27ac) it can be more difficult to relate sequence differences to activity differences. It is easier for transcription factors.
04:33:37Christopher Tuggle:@BMurdoch- have you developed plans how you plan to display these annotations on the new genome assembly- for browsers?
04:33:54Dominique ROCHA:@Brenda: (sorry a question not directly related to your talk) currently no public databases (ENSEMBL, UCSC, iSheep...) contain genetic variants (SNPs/indels) located on the ovine mitochondrial genome. :(( Do you know if this missing info will be added soon or where it can be found?
04:36:18Dominique ROCHA:@Brenda: OK thanks. I'll contact you. :)
04:38:07Peter Harrison:@Dominique Really interesting question. I will feed that back into Ensembl and see if there are any thoughts/plans on this. Remind me the next time we are having a chat on BovReg or EuroFAANG!
04:38:40Brenda Murdoch:Thank you Peter!
04:38:59Dominique ROCHA:@Peter: OK but I might forget to remind it. :(
04:41:08Peter Harrison:I know that Ensembl will be updating sheep this year so it is a good moment to look into this.
04:43:57michael sussman:How do you handle metadata associated with the features. Is there compression and encoding?
04:49:04Dailu Guan:IGV is a local software. If we load many samples, whether we need a machine with high memory? Have you tried how many samples you can load simultaneously?
04:51:06Peter Harrison:@Ted Really interesting points Ted, think this is an important next phase for FAANGs data reusability, ensuring that later stage analysis data is shared and can be interrogated. Interactive cloud based analysis platforms that lower the barrier to groups analysing data such as use of BigQuery in google cloud platform, ensuring integration of tracks into Ensembl (new brokering submission tool coming soon from data portal) and UCSC and more resources such as FAANGmine would really help our community. EuroFAANG infrastructure proposal will be looking into these barriers in collaboration with Elixir and The European Open Science Cloud (EOSC).
04:52:01Guillaume Devailly:@Ted Kalbfleisch in addition to .bam files, are you providing gene/transcript expression tables? I found those very usefull too for quick exploration analyses.
// end //