I want to use A clustering to find DMRs, with Aclust package. My problem is that Aclust uses IlluminaHumanMethylation450k.db package, which is deprecated, therefore when I'm trying to use a command:
ordered = order.betas.by.chrom.location(bVals)
I get following error (partially it's in Polish, but the most important part is visivle I believe):
Error in as.list(IlluminaHumanMethylation450kCHR37[probe.vec]) :
błąd podczas obliczania argumentu 'x' przy wybieraniu metody dla funkcji 'as.list': Error: object 'IlluminaHumanMethylation450kCHR37' not found
Does anyone have some ideas, advices how to solve this problem? I am quite determined to use Aclust, I know there is also Python version, but I assume there may be similar problems.
This support site is for Bioconductor packages. The Aclust package seems to be sort of orphaned. I don't see that it even exists on Tamar Sofer's original website, and the only version around seems to be what Pete Hickey has in his GitHub repo. But since it isn't a Bioc package anyway, this is not the place to be asking for help.
You can
Talk to Pete directly and see if he has some workaround that he is using.
Talk to Tamar Sofer and ask him what's up.
Clone Pete's GitHub repo and make it work using the FDb.InfiniumMethylation.hg19 package instead of the (deprecated and busted besides) IlluminaHumanMethylation450k.db package. That might be a fun project.
Use a sufficiently old version of R/Bioconductor that will still install the IlluminaHumanMethylation450k.db package. That might be the easiest thing to do, but it's probably also the most boring as well.
Jim's covered your main options. Just adding that I haven't tried to use Aclust in quite some time and don't have a version with proper 450k annotation support nor do I plan to make one.
It sounds like you've come across this, but in case you haven't, Brent Pedersen re-implemented the Aclust algorithm in Python (https://github.com/brentp/aclust). He also wrote the Python software, 'crystal', that integrates with his version of aclust for downstream associated testing of the clusters with phenotype (https://github.com/brentp/crystal). Since these are written in Python, they obviously don't directly depend on the IlluminaHumanMethylation450k.db package. So you might also have
5. Use aclust/crystal Python software to perform a similar analysis
This comes with the proviso that it looks like Brent is no longer actively developing aclust/crystal - but then it sounds like the original Aclust is somewhat abandoned and Brent is generally a very responsive and helpful developer.
Jim's covered your main options. Just adding that I haven't tried to use Aclust in quite some time and don't have a version with proper 450k annotation support nor do I plan to make one.
It sounds like you've come across this, but in case you haven't, Brent Pedersen re-implemented the Aclust algorithm in Python (https://github.com/brentp/aclust). He also wrote the Python software, 'crystal', that integrates with his version of aclust for downstream associated testing of the clusters with phenotype (https://github.com/brentp/crystal). Since these are written in Python, they obviously don't directly depend on the IlluminaHumanMethylation450k.db package. So you might also have
5. Use aclust/crystal Python software to perform a similar analysis
This comes with the proviso that it looks like Brent is no longer actively developing aclust/crystal - but then it sounds like the original Aclust is somewhat abandoned and Brent is generally a very responsive and helpful developer.
Thank you very much! Answers from both of you were very useful - at leat I know what options do I have now. Thanks a lot once again!