search

camiolomj / ICLite

Other
1 stars 0 forks source link

Number of Row clusters exceeds numbers of rows. #1

Open mdmanurung opened 3 years ago

mdmanurung commented 3 years ago

Dear authors,

I am very interested to use this package for my dataset. I also have paired CyTOF + RNA-Seq data.

I ran the function and received the following error:

Error in blockcluster::coclusterBinary(test_mat, semisupervised = FALSE,  : 
  Number of Row clusters exceeds numbers of rows.

Could you enlighten me on the most likely cause of this error?

By any chance, do you have a more comprehensive vignette to use this package? Suggestions on parameter tuning and interpreting the results would be a great start.

Thanks in advance.

Best regards, Mikhael

camiolomj commented 3 years ago

Mikhael,

Thank you for your interest in ICLite! We are currently working on a STAR Protocol submission with an expanded vignette and appreciate your patience during the early days of the package.

Your input error seems to be related to either A) formatting of the input gene expression matrix (which may need to be transposed) or B) very high number of assumed clusters (which seems less likely). If you use the load_IMSA_data() function you can match the formatting style of the paper's data which should help.

When picking input parameters, I think it's best to start broadly. Something along the lines of minimum connectivity = c(50, 100, 200), rho cutoff = c(0.4, 0.45, 0.5) and clusters = (15, 25, 35). You will find that not all combinations generate successful blockclustering solutions and this may change slightly between runs if you do not set a global seed before running the wrapper.

For evaluating results, I would suggest starting with gene ontology enrichment analysis of the output gene lists. The cell to module correlation map should identify some overlaps. For our data set, we knew that T cell relationships were something of a "positive control" while developing the package. I would suggest looking for rational connectivity in functional relationships between ontology terms and cells as a starting point.

I hope this helps! Please write with further questions,

Matt

On Fri, Apr 16, 2021 at 1:31 PM manurungmd @.***> wrote:

Dear authors,

I am very interested to use this package for my dataset. I also have paired CyTOF + RNA-Seq data.

I ran the function and received the following error:

Error in blockcluster::coclusterBinary(test_mat, semisupervised = FALSE, : Number of Row clusters exceeds numbers of rows.

Could you enlighten me on the most likely cause of this error?

By any chance, do you have a more comprehensive vignette to use this package? Suggestions on parameter tuning and interpreting the results would be a great start.

Thanks in advance.

Best regards, Mikhael

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/camiolomj/ICLite/issues/1, or unsubscribe https://github.com/notifications/unsubscribe-auth/ASJ4KCLQB33SKLHIQTJZQH3TJBX5XANCNFSM43B5YFYQ .