A Salty Week

This week I spent a lot of time working with data collected on the water samples from Astoria.  After running the samples through the FlowCAM and eliminating the unwanted metrics, I performed principal component analysis (PCA), taking the first three principal components (PCs). 
I was very excited to see the results obtained from PCA on the FlowCAM data since we had a previously published paper to compare results to.  In our experiment, we increased the salinity of the samples by increments of 5 g/kg of salinity from 0 to 25 g/kg.  In the paper a smilier approach was taken and they observed increasing aggregation of particles.  At first glance, our data was not producing such results.  I was a little bit worried.
My next step was to find a way to graphically represent the PCA results.  After creating a table that correlated three principal components with each sample, the size of the marker a function of variance, it was much easier to interpret the data.
Upon first glance of the graph I became encouraged.  It was clear that as salinity increased, particle width became a principal component and it accounted for an increasing amount of the variance.
In my opinion, this was one of the first confirmations that our measurement processes and sample preparation methods were correct and able to produce reproducible results.  I was very pleased, and quite frankly pleased with the graphical representation of the PCA results.  It was previously very difficult to quickly interpret PCA results.  It just goes to show how important presentation and communication of data can be.  Sometimes it seems as though people gloss over that part of data preparation.  Without proper and efficient data presentation, science would advance slowly.