Closed jarbet closed 1 year ago
Hi @jarbet,
@luqingan is the author of that code: hopefully she can help clarify it.
Hi @jarbet,
Thanks for the inquiry. The issue should be resolved. And yes, IMPACC
only needs to be run once. You can construct clustering assignments using the IMPACC_cluster()
function with ConsensusMatrix
generated by IMPACC()
and K
of your choice.
Please let me know if you have any questions.
-Luqin
Hi @jarbet,
Thanks for the inquiry. The issue should be resolved. And yes,
IMPACC
only needs to be run once. You can construct clustering assignments using theIMPACC_cluster()
function withConsensusMatrix
generated byIMPACC()
andK
of your choice. Please let me know if you have any questions.-Luqin
Thanks Luqin! I see it works now if you pass impacc.fit$ConsensusMatrix
to the css
argument. At first I was not sure what the css
argument meant, perhaps it would be good to specify in the documentation that this in the ConsensusMatrix from IMPACC/MPCC, or add a working example?
data(mtcars);
set.seed(1234);
impacc = IMPACC(
d=t(mtcars),
reps = 100,
verbose = FALSE,
adaptiveFeature = FALSE
)
IMPACC_cluster(
css = impacc$ConsensusMatrix,
K = 3
);
Also, I notice the yan
dataset no longer exists:
> library(IMPACC)
> data(yan)
Warning message:
In data(yan) : data set ‘yan’ not found
Hi @jarbet,
Thanks for the comments! I added the yan dataset back and a tutorial to the homepage. And the css
argument in IMPACC_cluster
is changed to ConsensusMatrix
to avoid confusion.
Cheers, Luqin
hi @luqingan , I can't find the yan dataset to be able to run the example, could you tell me where to find it?
hi @bulaciox ,
Sorry, there were some issues with data on iCloud. I just updated it and should be able to access the data by data(yan)
now.
hi @bulaciox ,
Sorry, there were some issues with data on iCloud. I just updated it and should be able to access the data by
data(yan)
now.
After installing the latest version and restarting my R session, data(yan)
works now. However, now IMPACC_cluster
is missing again:
library(IMPACC);
data(yan);
impacc <- IMPACC(d=yan$sc_cnt,reps = 100,verbose=FALSE);
clus <- IMPACC_cluster(ConsensusMatrix = impacc$ConsensusMatrix,K = 3);
Error in IMPACC_cluster(ConsensusMatrix = impacc$ConsensusMatrix, K = 3) : could not find function "IMPACC_cluster"
hi @jarbet , i'm so sorry for the issues, it should work now.
hi @jarbet , i'm so sorry for the issues, it should work now.
Thanks! The tutorial works for me now.
Hi @jarbet and @luqingan: Thank you for working this out!
@jarbet: I'm glad you're trying out IMPACC
and I hope it works well for you. @luqingan has gotten some fantastic results with it and we're excited to see it perform "in the field." Let us know if you run into any more trouble or if we can be of further help.
Cheers!
Hi @jarbet and @luqingan: Thank you for working this out!
@jarbet: I'm glad you're trying out
IMPACC
and I hope it works well for you. @luqingan has gotten some fantastic results with it and we're excited to see it perform "in the field." Let us know if you run into any more trouble or if we can be of further help.Cheers!
Hi @michaelweylandt: thanks, I am enjoying using IMPACC
so far as it is significantly faster than other consensus clustering packages I've tried, and it is giving me good results so far. I actually found out about IMPACC
from Dr. Genevera Allen's talk at the Lange Symposium, which was a fantastic talk!
Cheers!
Hi @luqingan,
Just a minor issue but wanted to let you know that the help page for IMPACC_cluster
still shows the css
argument, but this argument was changed to ConsensusMatrix
. The github README tutorial code works for me, I think the help page just needs to update the argument name.
Thanks
Although the help page
?IMPACC_cluster
works, when I try to useIMPACC_cluster()
I get the following error:Lastly, is the purpose of this function such that I can run
IMPACC()
only one time, and then useIMPACC_cluster()
to quickly get results for different values ofK
? I think in theory the main IMPACC algorithm only needs to be run once, and then you should be able to quickly post-process results to get cluster assignments for varyingK
, right?