Closed UK: DecodeME updates, was recruitment thread.

[MrMagoo]

Excited!

 

[hotblack]

May be worth reminding everyone of the blog post on the analysis plan.

And here’s a direct link to the delicious pdf with all the details.

I’m not going to pretend I understand it all, but there were bits I understand a bit more about now, compared to when the project started, thanks to my time on these forums.

 

[hotblack]

I’ve been re-listening to some older presentations on DecodeME and going through the analysis plan in more detail and trying to learn more about the processes involved to prepare my brain for whatever comes.

Here’s a few thoughts, hey maybe interesting to others and maybe people with more experience can chip in and answer any questions. Apologies if these are answered elsewhere.

Firstly it just seems incredibly rigorous. Really thorough and deep QC process for the data. Is this normal in all GWAS studies or extra mile stuff? It doesn’t surprise me given the team but still seems impressive.

The imputation steps seem interesting. Would it be fair to say this increases the overall number of variants available to look at? Do we have an idea of the number or scale of increase this will give us? This looks like it would disproportionately increase the resolution in specific areas (of interest) like HLA and mitochondrial dna, is that right?

Replication with other another group is mentioned, is this taking place now or later? Do we have timelines for fingenn?

Will we get the main GWAS results first and then see post GWAS steps taking place later, or will we see everything mentioned in the analysis plan all at once?

 

[V.R.T.]

Replication with other another group is mentioned, is this taking place now or later?

I reckon this is probably the Lipkin study that had funding pulled. Some of the US institutions seem to have won a court case to reinstate funding so maybe that will go ahead eventually.

 

[hotblack]

I reckon this is probably the Lipkin study that had funding pulled. Some of the US institutions seem to have won a court case to reinstate funding so maybe that will go ahead eventually.

Maybe there are mutliple groups I’m not sure. But fingenn is specifically mentioned and they seem to have generic data already, but I don’t know about this bit of the analysis. It could even be concurrent.

 

[V.R.T.]

Maybe there are mutliple groups I’m not sure. But fingenn is specifically mentioned and they seem to have generic data already, but I don’t know about this bit of the analysis. It could even be concurrent.

Oh this is very interesting, i didn't know about this.

Edit: it says in the link that Finngen results are now 'publically available for the whole research community'. So it's entirely possible it's concurrent.

 

[wigglethemouse]

Will we get the main GWAS results first and then see post GWAS steps taking place later, or will we see everything mentioned in the analysis plan all at once?

I assumed the GWAS results would be all at once as Chris Ponting stated in a webinar that funding runs out in August. Follow up studies using the data generated such as the announced projects by Precision Life and University of Glasgow will follow.

 

[hotblack]

I assumed the GWAS results would be all at once as Chris Ponting stated in a webinar that funding runs out in August.

After looking through the plan in more detail I wasn’t sure if we’d see all the steps outlined in section 6 (including things like PheWAS) or get the initial base GWAS results first then followups with the extra detail. It looks like there’s potentially a lot of different and interesting pieces here, not just one result. But you’re probably right and they’ll want/need to finish everything outlined at once.

Edit: PrecisionLife already have the raw genetic data don’t they? They’re not building on top of the GWAS, or at least that was the impression I got from the Genetics Centre of Excellence webinar a couple of weeks ago?

 

[wigglethemouse]

Ron Davis holds the yearly Stanford closed working group meeting with 80+ attendees in early September. I don't know the date and the meeting is not public.

The Stanford meeting is focused on how the key findings presented by researchers (some pre-publication, hence closed meeting) link to other research to get scientists thinking. talking, and asking questions about mechanisms that are core to the disease and can explain the patient experience.

Chris Ponting presented last year and I'm hoping he will present the results of Decode ME this year. This will be a great opportunity for Chris to get others thinking and talking deeper on how DecodeME results tie in to the biochemical processes of the disease (I hope).

@Simon M @Andy Do you know if Chris Ponting will be attending the Stanford meeting. Might be an opportunity for him to discuss with Ron Davis beforehand how to encourage a deep discussion on the DecodeME results and what's needed next.

 

[Andy]

Do you know if Chris Ponting will be attending the Stanford meeting.

No, I don't know, and honestly I would be surprised if he knew himself, given how much is going on at the moment.

 

[hotblack]

I have been thinking about the data analysis plan again in light of recent discussion of studies focussing on more severely affected people.

There are plans for all sorts of analysis, including stratified analysis for things like sex, co-occurring conditions, onset, etc. but there’s no mention of doing this for severity.

As far as I can tell there are around 2000 severely affected in the DecodeME cohort. Would that be enough to tell us anything?

With the 5:1 ratio of female to male that would mean around 3000 males (although it may be fewer as they say being female correlates with greater severity) and if performing stratified analysis with that group perhaps 2000 is enough to be of some use?

Does anyone have the proper breakdowns rather than my guesses? Or know if stratified analysis by severity is on the cards?

 

[hotblack]

To partially answer my own post, the data analysis plan says they can perform analysis on co-occurring conditions with “stratum of at least 1000 samples”. So the mild/moderate vs severe comparison may be possible. It would be really interesting to know if any genetic signal is strengthened or weakened here I think.

 

[Jesse]

I have been thinking about the data analysis plan again in light of recent discussion of studies focussing on more severely affected people.

There are plans for all sorts of analysis, including stratified analysis for things like sex, co-occurring conditions, onset, etc. but there’s no mention of doing this for severity.

As far as I can tell there are around 2000 severely affected in the DecodeME cohort. Would that be enough to tell us anything?

I asked the same thing in that thread:

Will DecodeME be able to classify different groups? For example based on severity or finding sub-types? Will it have enough statistical power?

This is what @Jonathan Edwards noted:

I suspect that once we have data from DecodeME it will be relatively straightforward to crosscheck with much smaller cohorts in away that allows statistically valid subgrouping. It may be difficult to torture more information out of DecodeME data per se.

Anyways I also really hope we can look at this one way or another!