You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Maybe this isn't the typical case for most users, but my experiment has multiple replicates of treatment and control. If I wanted to check how concordant my replicates were, and how separated my treatment from control conditions were, I'd likely attempt a PCA based on the raw or normalized counts produced by MAGeCK, coloring the points both by sample and condition (similar to DESeq2 output in nf-core/rnaseq). This could also help identify whether any sample(s) were outliers compared to others, which would be supported by the countsummary table--ie perhaps one sample was an outlier from others in PC space and also has a lower-than-usual mapping rate or very high Gini index etc.
I'd also plot the cross-sample pairwise correlations of normalized counts as a (n * n) heatmap, which likely also would be a useful output
The text was updated successfully, but these errors were encountered:
Description of feature
Maybe this isn't the typical case for most users, but my experiment has multiple replicates of treatment and control. If I wanted to check how concordant my replicates were, and how separated my treatment from control conditions were, I'd likely attempt a PCA based on the raw or normalized counts produced by MAGeCK, coloring the points both by sample and condition (similar to
DESeq2
output in nf-core/rnaseq). This could also help identify whether any sample(s) were outliers compared to others, which would be supported by thecountsummary
table--ie perhaps one sample was an outlier from others in PC space and also has a lower-than-usual mapping rate or very high Gini index etc.I'd also plot the cross-sample pairwise correlations of normalized counts as a (
n * n
) heatmap, which likely also would be a useful outputThe text was updated successfully, but these errors were encountered: