What are the expression value of genes/probes in limma results of topTable
1
0
Entering edit mode
rkp • 0
@rkp-10175
Last seen 3.6 years ago

Dear All,

Kindly suggest me which value are expression value in the following table that can be utilized for plotting Heatmap.

logFC   AveExpr         t      P.Value    adj.P.Val        B
-6.828181  7.966478 -18.82087 5.665531e-14 5.032385e-10 21.33399
4.855277 10.119313  18.69994 6.384443e-14 5.032385e-10 21.23390
-3.855238  9.043565 -18.66360 6.618656e-14 5.032385e-10 21.20367
-6.228212  4.886173 -17.73280 1.702977e-13 9.711228e-10 20.40300
-8.334742  8.435286 -16.80577 4.565306e-13 2.082693e-09 19.55247
-6.687395  6.121391 -16.62751 5.548775e-13 2.109459e-09 19.38246
-4.079570 10.274357 -16.09434 1.005614e-12 3.276864e-09 18.86092
-6.300424  7.211431 -15.17983 2.904778e-12 7.504834e-09 17.91840
-5.222531  6.654759 -15.16368 2.961136e-12 7.504834e-09 17.90119
-7.688590  9.899726 -14.68290 5.290129e-12 1.206678e-08 17.37925

Heatmap • 593 views
0
Entering edit mode

You normally use the normalised matrix and select probes/genes of interest from a statistical test, unless the variance in your data is very strongly related to your variables of interest - in this case you would use the most variables probes/genes and cluster these. I find aheatmap from the NMF package particularly good for this.

1
Entering edit mode
Aaron Lun ★ 27k
@alun
Last seen 10 hours ago
The city by the bay

None of the above. There's not much point making a heatmap from a single set of values - see the discussion at A: Heatmaps with replicates or triplicate data sets with counts to EdgeR, DESeq, or. If you want to make a heatmap, I would suggest using the original matrix of expression values. If you're using a one-way layout without an intercept, you could also use the fit\$coefficients to visualise the average expression of each group (assuming you have more than two groups, otherwise the heatmap would be fairly boring).