Having trouble creating a heatmap in SPOTFIRE (microarray data)

Question

[madkitty]

I'm trying to create a very simple heatmap from a dataset with no transformation.

Our data table was imported successfully and was defined as follow with 600 genes and 30 cell type (numbered from 1 to 30), the values correspond to gene expression values from microarray data :

Gene name     Cell1     Cell2
Gene1         120       100.58
Gene2         12.25     1000
Gene3         0.2       10

When creating the heatmap, I followed instruction on spotfire website, click on insert > New visualization > Heat Map

Then, on the Y-axis I select the gene names, from here I faced 2 major issues:

1. The heatmap displays only one cell type on the X-Axis, when I click on add, it adds two types of labels, and aggregates with an average value. so 30 cell types turns into 60 columns.
2. I just want to use the values of the dataset, without any aggregations, I don't understand why the aggregation shows up at the bottom of the heatmap and why I can't turn it off? It seems that the option "none" is missing in the drop down menu of the aggregation.

Is there a simple way to just have the gene expression vs cell 1 to 30 on a heatmap in spotfire? (So we should have 600 rows and 30 columns)

Answer 1

[Ahill]

If you want a clustered heatmap, you can use the short-wide table in the format you show and the Hierarchical Clustering tool, as mentioned by swbarnes2 above.

Recent Spotfire versions expect a tall-narrow (unpivoted) data table for the standard heat map visualization. If that's what you want, you can unpivot your data outside or inside Spotfire. To do it inside Spotfire,

File > Add Data Table > From (your existing table).

Select Unpivot transformation.

In Unpivot, select Gene column as Passthrough.

Select all Cell columns as Columns to Transform.

Click OK to add the unpivoted tall-narrow table. It will have 3 columns: Gene, Category (Cell), and Value.

Once you have the tall-narrow format data table, add your heat map.

Insert > New Visualization > Heat Map, and visualize the new unpivoted data table.

Set the row selector to Gene; the column selector to Category (Cell).

You'll get a heat map with Genes as rows and Category (Cell type) as the columns.

[madkitty]

Wow that really worked like a charm! Though when I display the heatmap it sorted out the category (cell type) by alphabetic order, is there a way to have no sorting at all?

Thanks a lot XD

Answer 2

[swbarnes2]

You probably should say what version of spotfire you are using. I'm using 4.0, and that's not how I make heatmaps. I use "Hierarchial clustering", and then it outputs a heatmap.

[madkitty]

I'm using version 7.0 :)