Question: Plot genes with similar expression patterns
0
10 days ago by
ecg1g150
ecg1g150 wrote:

Hi,

I have been using the "Analyzing RNA-seq data with DESeq2" (Love, Anders and Huber) and worked really well so far, thanks guys!

However, I would like to export and plot the different expression patterns of genes within my dataset. What I mean is to be able to group them by similar expression patterns +/- X error and see expression similarities responding to the conditions, and have different expression "trends" within the transcriptome.

I was wondering if there would be any function on DESeq2 or any other package that could do something similar? It would be interesting!

Cheers,

rnaseq deseq2 R • 76 views
modified 10 days ago by Michael Love23k • written 10 days ago by ecg1g150

why not trying to cluster them based on their expression (heatmap, kmeans, SOM etc.) and than plot the separate groups. DESeq(2) was not meant for such results.

Answer: Plot genes with similar expression patterns
0
10 days ago by
Michael Love23k
United States
Michael Love23k wrote:

I would start by computing the VST:

vsd <- vst(dds)


And then you can perform hierarchical clustering after computing Euclidean distance between genes. You can remove the row mean from each row, so that the baseline expression doesn't factor in.

Thanks Michael, I have done many types of clustering using vsd. But more than doing a hierarchical clustering, I mean to plot the actual expression profiles (y-axis) over time/geography (x-axis), and then generate the cluster plots by profile similarity. (ie Cluster 1: gene expression gradual increase, cluster 2: gradual decrease; 3, peak in the middle, etc)

I have found the package "Mfuzz" from Bioconductor that apparently is available to work with RNASeq. Have you heard of it? Do you recommend any other?

Cheers

Go for it. I have made these plots manually one time using base R: hclust, dist, rowMeans, lines, etc., but we don't have example code for this in the vignette.

Thanks Michael, I have done many types of clustering using vsd. But more than doing a hierarchical clustering, I mean to plot the actual expression profiles (y-axis) over time/geography (x-axis), and then generate the cluster plots by profile similarity. (ie Cluster 1: gene expression gradual increase, cluster 2: gradual decrease; 3, peak in the middle, etc)

I have found the package "Mfuzz" from Bioconductor that apparently is available to work with RNASeq. Have you heard of it? Do you recommend any other?

Cheers

I think the maSigPro package would be the one that you are looking for. I hope this helps.