Kaitlyn’s notebook: ASCA results

I used the new silo3_9 table I made for kmeans clustering and modified Shelly’s code to do the ASCA. Factor 1 is temperature and factor 2 is time.

Helpful tips:

ASCA.calculate()
ASCA-results-silo3_9

A combination of variables can explain the total variation based on temperature (PC1 = 100% for temperature) while 52.15% of the variance is explained by a single principal component for time. The interaction of time and temp produce the least amount of explained variance for PC1 (42.20%).

The sum of squares model describes how well the data fits a linear regression. Centering the data is important because PCA is a regression model without intercept. Non-centered data can be misleading since the eigen vector may point in the appropriate direction. Time contributes to the variance the most at 63.84% if the data is centered.

Temperature loadings plot of PC1 and PC2:
temp-loadings-plot

Time loadings plot of PC1 and PC2:
time-loadings-plot

Interaction loadings plot of PC1 and PC2:
interaction-loadings-plot

Single Score plot for Temp:
temp-scores-plot

Single score plot for Time:
time-scores-plot

Single score plot for interaction:
interaction-scores-plot

Score plot with projected data for Temp:
temp-projected_scores-plot

Score plot with projected data for Time:
time-projected_scores-plot

Score plot with projected data for Interaction:
interaction-projected_scores-plot

ASCA.DoPermutationTest() shows that no factors have a significant effect on protein abundance, but time has the greatest effect.
1(Temp): 0.119
2(Time): 0.015
12(Interaction): 0.191