Question: How to extract the result table from NMF clustering (R package 'NMF')?
0
gravatar for yuezhao97
5 weeks ago by
yuezhao970
yuezhao970 wrote:

Hi biostars,

I'm currently doing the analysis on my RNA-seq data of ~100 cancer samples and want to cluster them. I used the NMF clustering (run times = 200) and got a consensus cluster plot. Now I want to dig deeper into my results but can't find where to get the result table telling me which sample belong to which cluster. I'm wondering if anyone could help me with extracting the result table from the NMF clustering? Any information is appreciated. Many thanks!

rna-seq • 217 views
ADD COMMENTlink modified 5 weeks ago by Jean-Karim Heriche19k • written 5 weeks ago by yuezhao970
2
gravatar for Jean-Karim Heriche
5 weeks ago by
EMBL Heidelberg, Germany
Jean-Karim Heriche19k wrote:

For result <- nmf(...), the matrices are in result@fit@W and result@fit@H and can also be extracted with W <- basis(result) and H <- coef(result).

ADD COMMENTlink written 5 weeks ago by Jean-Karim Heriche19k

Thank you, Jean-Karim Heriche, for your kind and useful information.

Just one more question, I ran the basis(result) and coef(result), and they gave me two matrices consisting of the values for genes and samples for clustering. Should I pick the largest value for each sample to determine the right cluster it belongs to?

ADD REPLYlink written 5 weeks ago by yuezhao970

NMF can be seen as a form of soft clustering in which the values can be interpreted as cluster membership "strength" or relevance. By picking the largest value, you're performing hard clustering, assigning items to only one cluster.

ADD REPLYlink written 5 weeks ago by Jean-Karim Heriche19k

OK I see. Thank you so much!

ADD REPLYlink written 5 weeks ago by yuezhao970
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1314 users visited in the last hour