Have questions? Visit https://www.reddit.com/r/SNPedia

User talk:Donwulff

From SNPedia

great edits. Welcome to SNPedia. --- cariaso 18:52, 8 April 2014 (UTC)

please do not assign {{Topic}} so much. It is used to create an extra section in promethease reports which groups related SNPs. See Category:Topic for examples.

Aha, so it's kinda synonym for phenotype/(trait). I read the Category but it wasn't clear on what it's for; I guess I was looking for Glossary though there should probably be a Category for that.

gs286 looks correct, however be aware that those red rs#s will have to be created before Promethease can match the criteria. --- cariaso 07:27, 13 April 2014 (UTC)

re:big files. the wiki isn't the right place to host them. you can email them to cariaso@promethease.com and I will consider next steps. --- cariaso 16:28, 2 June 2014 (UTC)



From http://www.news-medical.net/news/20140925/e28098Missinge28099-schizophrenia-heritability-found.aspx referring to [PMID 25219520]:

The combined individual effects of 2891 SNPs previously associated with schizophrenia accounted for just 24% of the 
liability to schizophrenia in the cohort – 4196 cases and 3827 controls in the Molecular Genetics of Schizophrenia study.
But when considered as sets, the SNPs accounted for 90% of schizophrenia cases, report Igor Zwir (Washington University School of Medicine, St Louis, USA) and team.
“Our purely data-driven analysis shows that the elusive heritability of schizophrenia is not missing, but is encoded in 
a complex distribution of genotypic-phenotypic relationships”, they write in The American Journal of Psychiatry.

Which is kinda as I'd expected for complex phenotypes. What's cool and encouraging about this is they seem to now have a hang on the kinds of relationships between the variations to look for, and the computational power to tackle it even for relatively complex cases. Hopefully this can be successfully applied to other kinds of phenotypes, too. Although I'm still wondering how much is epigenetic, schizophrenia has usually been used as an example of epigenetics, I'm wondering if that's just been misattributed due to incomplete understanding.

This kind of got interesting fast, with "ten leading psychological genetics" refuting the claims of the scizophrenia study, via an online genetics blog at http://genomesunzipped.org/2014/09/eight-types-of-schizophrenia-not-so-fast.php . For their part, the scizophrenia study authors point out they've released their study through peer-reviewed channels, while the critics have chosen to ignore peer-review process: http://io9.com/geneticists-slam-study-claiming-schizophrenia-is-eight-1641303100 . I think we're also seeing a clash of the old way of publishing and online blogs in this issue.


"Biased Gene Conversion Skews Allele Frequencies in Human Populations, Increasing the Disease Burden of Recessive Alleles"

http://www.cell.com/ajhg/abstract/S0002-9297(14)00391-7

http://www.genomeweb.com/sequencing/study-suggests-biased-gene-conversion-can-ratchet-homozygosity-recessive-disease

This seems relevant and very troublesome for phasing, verification, imputation, haplotyping, biogeographic analysis and all sorts of rare-genotype work.


If we sort references, should they be in chronological order, or reverse chronological order? Right now new references seem to get added at end, so they're sorta-chronological, but for a "high-throughput" site like SNPedia I think showing the latest progress first would be more natural. Of course, IMNSHO the only right way to do this is to have the references, along with all the info-boxes on the pages, come from a database and not be part of the page itself, allowing sorting and filtering the references as one wishes. However, in the meanwhile they have to be in some order to allow seeing the natural development of the understanding and finding the latest on a glance. --Donwulff (talk) 02:54, 13 April 2014 (UTC)

In a perfect world I'd like newer stuff at the top, but autogenerated bot PMID titles don't belong above human generated summaries. So then the bot would have to be smart about where to insert it. And the bot isn't that smart, and probably never will be. Instead the solution has been bot populated stuff is appended to the bottom as it is found, while human summaries are free to claim the top of the page and reorder however seems best. --- cariaso 07:30, 13 April 2014‎ (UTC)
Okay, thanks. I've been doing both though, the human summaries and a list of references, because it's nice to see the original study title as well as chronological order of the study. The former, at least, would be helped if PMID's displayed title on mouse-over. Yeah, I know, it's a lot of work :) --Donwulff (talk) 07:42, 13 April 2014 (UTC)

I appreciate the spirit of your edit, but for practical reasons I've rolled it back. 99% of new papers are added by a bot, and it's not smart enough to maintain the ordering you want. Please stick with human summaries at the top, and papers in old->new chronological ordering below. If we ordered as you're suggesting, the bot would have to be able to distinguish human summaries from automatic content, and insert new papers into the middle. 'middle' is a somewhat abstract concept. bottom is very simple. I have re-added your new paper. --- cariaso 00:08, 22 January 2016 (UTC)

I don't actually have a strong preference to whether it's old->new or new->old, but I find that the completely random order they currently take it highly troublesome, since new studies frequently supersede the old ones, so you want to be able to read them in chronological order whether that order is from down to up or up to down. That is kind of a summary in itself. Moved earlier discussion we had on this from Talk:Rs1042838 into above so they're in one place. It's also not a huge problem if the bot keeps adding them to the end, but I can understand how it could lead to needless shuffling around if they're manually ordered every now and then. Best of all worlds would be if the references were managed automatically in a dynamic table allowing custom ordering and filtering, but yeah... However, it would be okay if I order them old to new? Donwulff (talk)
I'm flexible on this, but you should be aware that the bot is stupid, and factor it into you planning. I'd suggest, human summary of the big picture at the top. articles in roughly chronological ordering below. When there is a relationship between 2 articles feel free to group them together, and perhaps offer some human text nearby to explain the connection and summarize the combined conclusion. --- cariaso 23:42, 1 February 2016 (UTC)