*Thread Reply:* I used my university email this time instead of my gmail, hopefully it should reach people now. Those who’s junk it ended up in all had uni emails
*Thread Reply:* Hopefully that helps
*Thread Reply:* Although it is ridiculous that I get more 'spam' from institutional emails then Gmail 😂
*Thread Reply:* Hopefully! If not, I hope people check the website
*Thread Reply:* Ooh we should tweet too!
*Thread Reply:* Completely forgot
*Thread Reply:* I'm on my phone with Maia at the park, maybe you could draft one?
*Thread Reply:* Yes, please do and we can retweet
*Thread Reply:* Uhh ok I can also try
*Thread Reply:* Done: https://twitter.com/jafellowsyates/status/1305528478134132736?s=19
}
Twitter
*Thread Reply:* No problem, my mum took over
For those who are Twitter inclined: https://spaam-community.slack.com/archives/CPHECT30A/p1600097234008300?threadts=1600096599.006000&cid=CPHECT30A|https://spaam-community.slack.com/archives/CPHECT30A/p1600097234008300?threadts=1600096599.006000&cid=CPHECT30A
}
*Thread Reply:* Will get to that after Irina's initial discussions
*Thread Reply:* Probably discussion points 2
*Thread Reply:* that's what I used to do as well
*Thread Reply:* You'd think ;). But the reason why we ended up doing that because we had leaking of capture libraries into blanks
*Thread Reply:* And people got very Paranoid
*Thread Reply:* Ah, makes sense. With in-line barcodes that's not a problem. They have other problems, though, like reduced ligation efficiency depending on the barcode sequence
*Thread Reply:* I have, but still feel very fuzy about them!
*Thread Reply:* Awesome thanks! Could you please link the paper?
*Thread Reply:* I think the barcode bias discussion is in the supp
*Thread Reply:* Yes, we had to move it out of the main text so it's completely in the Supp Info document
Related to the genotyping and the issue of using just a single reference genome, there is this paper from last week that propose a nice turnarround by mapping into graphs rather than genomes... https://pubmed.ncbi.nlm.nih.gov/32943086/
*Thread Reply:* I was going to ask a question about this very paper 🙂
I think this is the Hungarian mummy paper: https://www.nature.com/articles/ncomms7717
*Thread Reply:* There's four modern genomes available
*Thread Reply:* As I know, there are only 4 modern RefSeqs of M. oralis, and maybe two modern calculus samples which contain well-covered M. oralis. These modern samples are all from @velsko or Mann et al. study.
*Thread Reply:* Thinking in terms of the variation graphs what would be good test data. Pathogen stuff obviously pretty good for that, but outside of that comes more difficult
*Thread Reply:* But I wasn't very confident in the results.
*Thread Reply:* Hi Sterling...why not confident?
*Thread Reply:* There were about 10,000 reads that mapped but because those references are mixed with soil, oral, skin and all sorts of microbes, I wasn't sure how to interpret it. If I were to go back, I think it may make more sense to use the HOMD database as a reference for MapDamage. Has anyone tried this yet with dental calculus?
*Thread Reply:* @Lena G @Pooja Swali what do you use?
*Thread Reply:* htsbox majority consensus caller by Heng Li so far, but haven't done a careful evaluation
*Thread Reply:* Oooh, not heard of that. What is the possible advantagE?
*Thread Reply:* Angsd, but I've only used it for single genes so far.
here is the very fresh & through review about strain lineation etc… https://pubmed.ncbi.nlm.nih.gov/32499497/
*Thread Reply:* I have a bit of experience with panX and also roary. Both easy to use but requires tweaking to incorporate ancient data
Anyone using DeepVariant ? https://github.com/google/deepvariant
*Thread Reply:* ```DeepVariant supports:
Germline variant-calling in diploid organisms. For somatic data or any other samples where the genotypes go beyond two copies of DNA, DeepVariant will not work out of the box because the only genotypes supported are hom-alt, het, and hom-ref. The models included with DeepVariant are only trained on human data. For other organisms, see the blog post on non-human variant-calling for some possible pitfalls and how to handle them.``` sigh
A nifty comparison of SNP calling pipelines for bacteria
*Thread Reply:* snp-sites works pretty well and extremely fast.
*Thread Reply:* And this one too https://github.com/Amine-Namouchi/snpToolkit
*Thread Reply:* how have you found snippy with aDNA?
*Thread Reply:* Haha I have only used snippy for making multi-sample SNP alignments because you must use their processing pipeline with little power to modify parameters. So I input bams that have been prepped using aDNA standards.
*Thread Reply:* yeah it is designed as a full pipeline so I was wondering how it does with aDNA bams and core snp alignments, but I guess we are back to lacking configuration options...
*Thread Reply:* A few of us in the poinar lab are using it like Kelly was saying
https://github.com/BoyanZhou/AntCaller
Here is a tool to detect recombinant region: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1004041
detecting recombination from metagenome data : https://www.nature.com/articles/s41592-018-0293-7
*Thread Reply:* http://megan.informatik.uni-tuebingen.de/
*Thread Reply:* There is a dedicated section for MALT
*Thread Reply:* Ok, thanks @James Fellows Yates I think I tried that one, did not get answers on my questions, just wanted to see if somebody in the workshop is actually a MALT developer
*Thread Reply:* Ah no, they aren't present here.
*Thread Reply:* But it's basically just Daniel who develops afaik
*Thread Reply:* Ok, thanks! Maybe I can throw my question to the community when touch MALT at this workshop?
*Thread Reply:* I would just like to comment on how wonderful the name gubbins is.
check slide 3 from here, we are going to make it public soon
*Thread Reply:* You mean regarding the lack of aDNA patterns?
*Thread Reply:* it might not be a universal solution but for certain situations might be useful. The stuff we are doing is inspired by this https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3799439/
*Thread Reply:* The only problem I could see is how to persuade the reviewer that is a non-standard, but valid approach. 😉
*Thread Reply:* microbes and their special needs 🙂
*Thread Reply:* there is a lot of info hidden in the coding space that we can exploit in combination with the methods of @Nikolay Oskolkov
*Thread Reply:* Definitely, there is a lot of reference data to train for codon usage.
*Thread Reply:* when we don’t have references, we use a super-read like approach to partition potential ancient reads
*Thread Reply:* This is in one of the slides for tomorrow’s talk
We also use this as an alternative to Mapdamage but is mainly useful for samples that are UDG treated https://github.com/pontussk/PMDtools
*Thread Reply:* Will ask next!
I would say you can compare but need to record the metadata, and do some statistical modelling taking them into account. Combat and Songbird are nice tool for that https://github.com/biocore/songbird https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-018-2263-6
*Thread Reply:* Agree, can’t compare 16s and WGS data for example, or metagenomic profiles generated by classifying agains different databases, or different tools. As an optimistic note: keeping track of the metadata and visualising their effect, on let’s say a PCA colored by batch effect, will give a good starting indication of whether or not this is something to worry about
*Thread Reply:* True! I usually make a correlation heatmap of my PCs against all available meta-data to understand the sources of variation in my data. If my PC1 is mostly explained by the batch variable, it is very worrying
Last one on batch effect correction for tonight: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1850-9
*Thread Reply:* But is it in this case ethical to work on that at all?
*Thread Reply:* I think it depends on the group.
*Thread Reply:* What is the motivation behind making the data not public?
Fort Lauderdale Agreement: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2672783/
*Thread Reply:* Especially museum samples collected along time ago, e.g. during colonial times are a delicate issue.
A note on radiocarbon reporting conventions
*Thread Reply:* @Maxime Borry has spoken with the person leading this but found she wasn’t very interested. Maybe having a group collectiveley approach would be more persuasive
*Thread Reply:* Yes, I can get in touch through the people from my former group in Bremen, when CAMI appeared they contacted us
*Thread Reply:* at that time was Alexander Sczyrba
*Thread Reply:* Well, let’s say that no one went forward with it.
*Thread Reply:* There is also the LEMMI benchmarking challenge that is quite interesting, because it is updated in real time https://lemmi.ezlab.org/#/ We contacted them, but it also didn’t go forward https://gitlab.com/ezlab/lemmi/-/issues/5
*Thread Reply:* This should be coupled with the lab standards @Christian Carøe was talking about. For example the guys from the XMP develop this https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5345951/ In last year GSC meeting Scott presented their standards
This is the paper Jamesis referring to: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1003531
*Thread Reply:* great paper! I had the same issue James (i.e. how to norm? etc.)
And this one: https://www.frontiersin.org/articles/10.3389/fmicb.2017.02224/full
*Thread Reply:* But then who should do the testing? I would disagree testing doesn't result in a publication..., 50% of bioinformatics is just hat 😆
*Thread Reply:* Bioinformaticians with permanent positions, e.g. as those offered by SciLifeLab
*Thread Reply:* Permanent positions?! They are a thing!?!?! 😉
*Thread Reply:* Positions in core-facilities are not necessarily permanent 😉
*Thread Reply:* Those at SciLifeLab are 🙂
*Thread Reply:* I was also referring to what I myself have an interest in - it's not so much whether I will get a publication out of it, it's also what I'm interesting in doing, which is more data analysis, not testing. But I agree testing is super important and useful!
*Thread Reply:* I think that testing is good and necessary, but it is necessary to put an end at some point, choose one solution we believe that will satisfy reviewers and community, and focus on the questions of the project. I personally care much more about questions than methods. So as long as the method is sounding, although not perfect, I prefer to move on
*Thread Reply:* I also find it interesting (@Katerina Guschanski), although you say testing will not result in pub, there is clearly a need for these testing papers as it seems one of the biggest concerns of people (particularly from the session yesterday). So if someone was to do that, and publish it, in principle they could get lots of citations?
*Thread Reply:* Absolutely true and a super good point. I/we would be happy to be able to use such guidelines. Also perfect arguments to provide to reviewers when they go "Why didn't you use this or that tool...?". So I guess the truth is somewhere in the middle. Periodic testing of new tools to compare to winners from the previous round?
*Thread Reply:* Exactly, if we had a CAMI-like dataset, so a 'gold standard' for people to test - would you feel more comfrotable for a student to spend time on it?
*Thread Reply:* In the sense there is already a precedence
*Thread Reply:* for this we need a way to generate ancient mock communities in the lab, no idea if this is a thing
*Thread Reply:* Nope, I don't think we could do that; unless we get a single Mammoth femur and freeze it
*Thread Reply:* But we would need stuff for different contexts; this isn't just microbes
*Thread Reply:* one step at a time
*Thread Reply:* That would be perfect (using a mock community for tool testing periodically). Ideally coordinated effort across labs with multiple participating researchers.
*Thread Reply:* Wouldn't it be possible to produce a "simulated" community with typical aDNA traits: Damage, contamination, fragmentation patterns, etc. It is possible for hosts...
*Thread Reply:* Yeah I guess. IIRC CAMI had both synthetic dataset and also a 'real' mock community (with spiked in stuff)
*Thread Reply:* yep the real stuff is the interesting
*Thread Reply:* Exactly, we can't really model aDNA datasets very well atm
*Thread Reply:* the synthetic should not be a problem
*Thread Reply:* Well, we don't know the ground truth of a real aDNA dataset at the moment 🙂
*Thread Reply:* Yeah like Gargammel uses a Neanderthal damage profile... because everyone is analysing that...
*Thread Reply:* you, can use your sample patterns, but I think we need more realistic models
*Thread Reply:* @Nikolay Oskolkov I guess we could use some DL to generate more reliable syntcomms
*Thread Reply:* You mean even more complicated than what we currently think is already nightmare 🙂?
and the standard: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4511511/
*Thread Reply:* > Examples of activities that alone (without other contributions) do not qualify a contributor for authorship are acquisition of funding; general supervision of a research group or general administrative support; and writing assistance, technical editing, language editing, and proofreading. Those whose contributions do not justify authorship may be acknowledged individually or together as a group under a single heading (e.g. “Clinical Investigators” or “Participating Investigators”), and their contributions should be specified (e.g., “served as scientific advisors,” “critically reviewed the study proposal,” “collected data,” “provided and cared for study patients”, “participated in writing or technical editing of the manuscript”).
Also the CREDIT one: https://casrai.org/credit/
*Thread Reply:* > Acquisition of funding, collection of data, or general supervision of the research group alone does not constitute authorship.
*Thread Reply:* How did you estimate this? I disagree from my experience with them
*Thread Reply:* I think SRA may screen and mask the human reads
*Thread Reply:* Stripped before uploading to ENA
*Thread Reply:* \don't know 😆
*Thread Reply:* Or you have to apply to some ethical body that sometimes they come back to you and sometimes not. E.g. dbGAP.
*Thread Reply:* Second that!
*Thread Reply:* Because we can 'immortalise' libraries 😉 (according to the mantra of a certain JK)
*Thread Reply:* Hahaha! I admit to have used this argument in the early years...
*Thread Reply:* but if the data cannot be compared to anything else (from our conversation yesterday) is it really a bioarchiving of samples?
*Thread Reply:* It did come from the same source
*Thread Reply:* This is really interesting - thanks for asking!
*Thread Reply:* Depends on the sample; bigger problem is short read lengths
*Thread Reply:* not always, there are other alternatives than MAGs
*Thread Reply:* It worked well for us for some (highly abundant) species from historical dental calculus samples
*Thread Reply:* i work with very complicated samples, and is very challenging
*Thread Reply:* so the traditional approach doesn’t work
*Thread Reply:* this is like going back 10 years back in time in metagenomics
*Thread Reply:* How do you handle the aDNA damages during the the assembly?
*Thread Reply:* we are working on integrating this in the assembly process
*Thread Reply:* I will show briefly
*Thread Reply:* also not using de brujin approaches
*Thread Reply:* very true but it was targeted - not an open exploration. There are only a handful of papers fidning ancient MAGs
*Thread Reply:* One of the few recovering ancient MAGs by @Katerina Guschanski and @Jaelle Brealey https://academic.oup.com/mbe/advance-article/doi/10.1093/molbev/msaa135/5848415
*Thread Reply:* The limitation is: these are very recent samples my most standards (within the last century) and the used approach worked for highly abundant species
*Thread Reply:* Exactly! Hence the need for the combination of assembly plus assembly free methods
*Thread Reply:* but it opens very exciting possibilities
*Thread Reply:* for example http://merenlab.org/2016/11/08/pangenomics-v2/
*Thread Reply:* we have a full workshop every year on this
*Thread Reply:* https://pagesperso.univ-brest.fr/~maignien/ebame5.html
*Thread Reply:* For Maxime and me, we mainly work on very well described datasets, like human gut or human oral microbiome, which makes it easier to compare to known data. For unknown environments, it is more complicated.
*Thread Reply:* @Antonio Fernandez-Guerra Is it too late to sign up for the workshop for this year?
*Thread Reply:* I would also be interested in attending this workshop
*Thread Reply:* it is an excellent workshop! But sadly i dont think it is happening this year.
*Thread Reply:* however keep your eyes on twitter or email Louis for further info for when they host it next: lois.maignien@univ-brest.fr
*Thread Reply:* Also bear in mind this is not geared for ancient metagenomes
*Thread Reply:* Thank you @Anna F.!
*Thread Reply:* there will be some online courses available early next year if that would be interesting? again not necessarily ancient
*Thread Reply:* I would still be itnerested. Are they hosted by the same people?
*Thread Reply:* partially! Tom Delmont will be on the teachers. It will be a short course so an intro to the field
*Thread Reply:* Next year unfortunately
*Thread Reply:* @Nikolay Oskolkov this one https://www.biorxiv.org/content/10.1101/490078v2 from Simon Rasmussen is using Variational Autoencoders
*Thread Reply:* CONCOCT from Chris Quince https://pubmed.ncbi.nlm.nih.gov/25218180/ works very nicely
*Thread Reply:* this last one from Chris is also super cool https://www.biorxiv.org/content/10.1101/2020.09.06.284828v1.full
@Maria Spyrou https://pubmed.ncbi.nlm.nih.gov/31235882/ PLASS
PenguiN: https://twitter.com/thesteinegger/status/1306245987048910849
}
Twitter
*Thread Reply:* What do you mean? Any channel, or something related to what we discussed today with the assembly of aDNA data? Antonio knows about it, I am just interested on it, but I'm still very ignorant on this respect
*Thread Reply:* I would add one like New computational methods or something similar, where we can discuss new approaches
*Thread Reply:* analysis-hipsters? ;)
*Thread Reply:* also one channel with something like Matchmaking for potential new collaborations
*Thread Reply:* You guys should be able to make these channels actually 🤔. Make them and advertise on #general! If you can't let me know
about unknowns… (shameless self promotion) http://merenlab.org/2020/07/01/dark-side/
The papers that describe the assembly methods used by the Segata lab are: https://www.sciencedirect.com/science/article/pii/S0092867419300017?via%3Dihub
https://www.sciencedirect.com/science/article/pii/S1931312819304275?via%3Dihub
*Thread Reply:* Yes, but very fragmented. ~40% region is confidently covered.
http://merenlab.org/2016/12/14/coverage-variation/
*Thread Reply:* https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5643016/ Luhmann 2017
*Thread Reply:* Segata lab poop: https://www.sciencedirect.com/science/article/pii/S1931312819304275?via%3Dihub
*Thread Reply:* Jaelle reindeer/bear/gorilla calculus: https://academic.oup.com/mbe/advance-article-abstract/doi/10.1093/molbev/msaa135/5848415
*Thread Reply:* hep B: https://elifesciences.org/articles/36666
*Thread Reply:* https://www.nature.com/articles/ncomms3172
*Thread Reply:* Segata lab poop: https://www.sciencedirect.com/science/article/pii/S1931312819304275?via%3Dihub
*Thread Reply:* Jaelle reindeer/bear/gorilla calculus: https://academic.oup.com/mbe/advance-article-abstract/doi/10.1093/molbev/msaa135/5848415
*Thread Reply:* assembly for y. pestis: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5643016/
*Thread Reply:* Assembly for ancient S. enterica Paratyphi C https://www.cell.com/current-biology/pdfExtended/S0960-9822(18)30694-8
*Thread Reply:* A couple from the lake sedaDNA world: https://www.biorxiv.org/content/10.1101/2020.04.10.035535v1 https://www.biorxiv.org/content/10.1101/2020.01.06.896068v1
*Thread Reply:* 😅 sorry I panicked as people were posting on different ones
*Thread Reply:* now there are several threads...
*Thread Reply:* 🤦 James fail
*Thread Reply:* I'll collcet them all in one place 😅
*Thread Reply:* Feel free to post (if you have time) specifci things you would like to have/see
*Thread Reply:* Hey, same for me here, I wont be able to make it to the last session. Thanks so much for organising and it was nice meeting you, also very interested to continue the communication in the future - I really like the idea of little workshops, maybe on assemblies :)
*Thread Reply:* Good to know! i’ve seen both to be honest - filtering out the host and not filtering out the host- it really seems to depend whose modern metagenomic papers you come across!
*Thread Reply:* I’m having the same issue with DIAMOND-produced sam files. I’d also like to know if anyone has a solution
*Thread Reply:* Hace you used the sparseSAM output? If so I would recommend turning that off
*Thread Reply:* If you didn't use that I need to check if I did anything special in a different project I was using them recently
*Thread Reply:* I don't think I did though
*Thread Reply:* No, the issue w/the DIAMOND sam files is that it’s a pseudo-sam file, since they aren’t designed for protein alignments. @Nikolay Oskolkov did you run MALT in BLASTX mode? Do you also have protein alignments?
*Thread Reply:* @ivelsko I ran it in BLASTN mode for aligning DNA reads. @James Fellows Yates I did not use sparseSAM, no, here is how I ran it:
malt-run -at SemiGlobal -ssc -m BlastN -i fastqfile -o fastqfile.rma6 -a fastqfile.sam -t 20 -d maltdatabase
*Thread Reply:* Hmm, ok that looks pretty similar to what I use but I think without the soft flipping (iirc that's what the SSC flag does?). I'm having a day off today, but if you ping me tomorrow I'll check to see what I did
*Thread Reply:* Was there any particular error samtools was reporting?
*Thread Reply:* Thanks @James Fellows Yates, you are right, the issue seems to be related to the soft-clipping. I will check without the -ssc flag. Still strange that if you want to implement soft-clipping, you get a strange sam-file that can not be read by samtools
*Thread Reply:* Yeah, there are a few wierd things about the malt implementation. I would still try posting on the megan forum again. Semester is starting up again so maybe the Devs are back at work. It took me one or two attempts to get replies previously
*Thread Reply:* yeah I don't see myself doing anything special in my code
*Thread Reply:* But I don't think I had to touch the cigar string
@channel Find here the draft of all the links/references posted during SPAAM2: https://docs.google.com/document/d/1oXS0MHg1DdB0xe3vlen5TEcC2SFZsp1aSwEOnK_soBM/edit?usp=sharing
Consider this a semi-annotated bibliography (I tried to put the context of each link, as far as I could remember/reconstruct from the chat), which we will bundle with the notes/slides from SPAAM2 and will make public.
Please feel free to make edits etc!
*Thread Reply:* Real time Rapid markdown fixes, love it 😂
Hi everyone Just a reminder that you have until this coming Friday October 16 to add your comments to the documents in the shared Google Drive folder: https://drive.google.com/drive/folders/1XPITeI2Sku9WScE85d4suyYXK1OyFgpU?usp=sharing
AND to fill out our survey: https://forms.gle/Z6mM9Zt9QXxHvk637
Thanks again for your contribution!
Best,
The Organization Team Standards, Precautions & Advances in Ancient Metagenomics 2
@channel the comments are now closed, thanks for those who left them!
I'm pleased to announce the slides/notes from our discussions are now all available (and citable 😉 under a CC-BY license with a Zenodo DOI! ) here: https://github.com/SPAAM-community/spaam-event-notes! Be creative how you use them to make new projects, presentations etc!
The feedback from the survey has also been sared with the <#C01BL1HD57T|spaam3-organisers> who will use it to make the next event even better 💪
*Thread Reply:* YEESSSSSSS ❤️❤️❤️❤️
*Thread Reply:* I am doing an nf-core/eager live demo at 13:00_13:30...
*Thread Reply:* (but could still do it, and I leave halfway through)
*Thread Reply:* @Clio Der Sarkissian I vote meeting at Pyrénées because how cool would it be to have a SPAAM meet up in the mountains!
*Thread Reply:* Not Cassoulet?
*Thread Reply:* OK YES FOOD
*Thread Reply:* (I thought it was a place)
*Thread Reply:* Holy shit that looks amazing
*Thread Reply:* Yes Cassoulet
*Thread Reply:* Pyrenees or Cassoulet? Pyrenees sounds more classy, but I’m not against a bit of fun.
*Thread Reply:* Did you also receive the e-mail to access your platform account?
*Thread Reply:* I thought we were supposed to get access in advance to fix texts so if you got it as well I might be working for nothing…
*Thread Reply:* Yes, I just logged in and that's how I found out about the different palces
*Thread Reply:* Well depends if you're feeling like gaining or burning energy 😆
*Thread Reply:* Bot hare good wit hme
*Thread Reply:* It is indeed a nice platform!
*Thread Reply:* aaargggh, OK, free time for me, not optimal English for us all, but let’s say it doesn’t matter as long as people understand
*Thread Reply:* Huh? Sorry I don't undersand...
*Thread Reply:* (ironicallly)
*Thread Reply:* thanks for explaining (not ironically)
*Thread Reply:* Oh wait, slack didn't tell me about htis message:
> I thought we were supposed to get access in advance to fix texts so if you got it as well I might be working for nothing…
*Thread Reply:* Everything is understandanble to me so far
*Thread Reply:* There is one page not translated but no-one would use it anyway
*Thread Reply:* Profile > A See my participant file
*Thread Reply:* But honestly 🤷
*Thread Reply:* You can work it out anyway
*Thread Reply:* Also noticed:
*Thread Reply:* In the chat box it's still french
*Thread Reply:* Then on hte 'online' page:
*Thread Reply:* But like I said, basically nothing people will ever use probably
*Thread Reply:* (or that they can't understand
*Thread Reply:* @Clio Der Sarkissian somethign that might be a bit more confusing:
On the right - when I started the private chat with Aida
*Thread Reply:* OK, thanks! I’ll let them know
@Clio Der Sarkissian is on TV!
*Thread Reply:* Starting with the good stuff I see, behind your laptop 💃
*Thread Reply:* That's entirely @aidanva 😂😂😂
*Thread Reply:* Sugars, scissors and a control remote… who can resist…
*Thread Reply:* Absolutely loved it Clio - really well done 👏👏👏
*Thread Reply:* No comments, but just wanted to say it was a great read, well don everyone. Table 2 is a great idea!
*Thread Reply:* Same ! It was a really nice recap for someone who just joined the field. Plus, I love figure 2 🤩!
*Thread Reply:* Thank you very much
