topic Re: JMP Genomics - Sample order in hierarchical clustering in Discussions

JMP Genomics - Sample order in hierarchical clustering

LouisAltamura — Sun, 11 Jun 2023 11:20:28 GMT

Under the ANOVA output for the Basic RNA-Seq Workflow, JMP Genomics will create a hierarchical clustering of both samples samples (columns) and the significantly differentially expressed genes (rows). However, is there a way to only cluster on the genes? I have a time series data set, and I would like the columns to be arranged by study time point for presentation purposes. I cannot find a way to accomplish this. Thanks in advance!

Re: JMP Genomics - Sample order in hierarchical clustering

Chris_Kirchberg — Tue, 14 Dec 2021 15:15:51 GMT

Hi @LouisAltamura,

You can cluster on Genes only and show the color map without clustering on samples, but one is not able to change the order of the columns after clustering. They would be ordered by the order in which they are put in the Hierarchical clustering dialog. The genes are also clustered based on those columns so changing the columns and trying to cluster again would change the clustering.

It sounds like you would like to keep the hierarchical tree of the genes, but use a different set of columns to create the heatmap and also to order the columns based on those columns information (time series in this case). If this is not the case, are you asking just to re-arrange the columns in the heatmap according to the time series in which they represent with out clustering the columns (two-way clustering?).

For example, are you looking to remove the clustering of the three samples (columns) below and re-arrange their order so that they represent a time series (like DOM, then CAS and then MUS)?

Re: JMP Genomics - Sample order in hierarchical clustering

LouisAltamura — Tue, 14 Dec 2021 15:50:10 GMT

Yes, I would like to keep the clustering of genes, but simply order the samples by timepoint. I don't want any clustering by sample.

The default hiearchical clustering graph ends up adding a lot of columns that I don't need for this presentation. So I have been relaunching the analysis and selecting only Std LSMeans columns that I need. Unfortunately, the process does not show the heat map by default in the order that I input the columns. I can only get the heatmap to appear when I perform a two way clustering.

Re: JMP Genomics - Sample order in hierarchical clustering

Chris_Kirchberg — Tue, 14 Dec 2021 16:09:48 GMT

Thanks for the update. It is possible to show the color map (heatmap) without two-way clustering. Under the red-triangle above the Two-way clustering option, there is an option called Color Map. Select that and then Choose.... Click on the the color scheme of choice and the heatmap will appear with the columns in order in which they were entered without having to do the two-way cluster. Unfortunately, the labels of the columns do not appear within JMP Genomics 10. It does appear in the next version of JMP (JMP 17) which is the version that will be used for the next version of JMP Genomics.

By the way, JMP 16 also will show the column names so if you have access to JMP 16, you could take the same resulting data set and cluster within JMP 16 instead.