Entering edit mode
6.7 years ago
1769mkc
★
1.2k
I plotted this correaltion heatmap using heatmaply i m getting cluster that forms good correlation. Is there any way i can take out the gene pairs or set of genes that are forming high correlation.
Second approach
I m making a correlation matrix . So is there a way to take out set of genes that are giving a good correlation.Such as if i set a threshold of lets say 0.6, is the threshold , how do i filter those genes. Any suggestion or help ,if i can do that in R?that would be helpful
What kind of down-stream analysis are you planning? What do you mean by 'take out', keep or filter, or remove redundant features for feature selection? If feature selection, there is a recipe here: http://machinelearningmastery.com/feature-selection-with-the-caret-r-package/ using the caret R package.
I would look for pathway with the genes that are showing really good correlation.
"What do you mean by 'take out', keep or filter, or remove redundant features for feature selection? " I meant that if there is a good correlation how do i take those set of genes and values?
What about your earlier question: Saving individual clusters from heatmaps which has a validated answer? please explain what is different with your new question.
I did that with complex-heatmap i took out clusters but i'm not sure if those clusters are highly correlated ,i did define cluster that the
km
parameter .So im not sure if those are highly correlated or not. So if they are forming a cluster then they are correlated ?