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
I am exploring this great tool! with my long-read data set.
When looking to get a more customized PCA plot, I encountered a calculation on the variance explained in BAMBU which differs from other results in different packages.
The way that the variance for PC1 and PC2 is calculated in BAMBU uses the output from prcomp command, taking into account the standard deviation like: round(pcs$sdev[1]/sum(pcs$sdev) * 100, 1)
In my case, I got a PC1= 8.1% and PC2 = 6.9% with BAMBU.
However, using the fviz_eig() function in the library(factoextra) which gives the contribution this number goes to 13% for PC1 and 10% for PC2 (please see attached figure).
This is also consistent with the output of summary (prcomp).
I am not a principal component expert, but I would like to understand why the calculation to obtain the variance in the PCA plot was calculated in this way. And maybe, if it is needed adjust the BAMBU code for reproducibility purposes.
All the best,
Niko
The text was updated successfully, but these errors were encountered:
Hi there,
I am exploring this great tool! with my long-read data set.
When looking to get a more customized PCA plot, I encountered a calculation on the variance explained in BAMBU which differs from other results in different packages.
The way that the variance for PC1 and PC2 is calculated in BAMBU uses the output from prcomp command, taking into account the standard deviation like:
round(pcs$sdev[1]/sum(pcs$sdev) * 100, 1)
In my case, I got a PC1= 8.1% and PC2 = 6.9% with BAMBU.
However, using the fviz_eig() function in the library(factoextra) which gives the contribution this number goes to 13% for PC1 and 10% for PC2 (please see attached figure).
This is also consistent with the output of summary (prcomp).
I am not a principal component expert, but I would like to understand why the calculation to obtain the variance in the PCA plot was calculated in this way. And maybe, if it is needed adjust the BAMBU code for reproducibility purposes.
All the best,
Niko
The text was updated successfully, but these errors were encountered: