Nice video, Dan. Would it be interesting to consider the non-independence of chemotaxis and flagella (unless chemotaxis specifically does not include flagella genes in this case) and see how the results come out using an interaction (CHEMOTAXIS:FLAGELLA) in the CCA calculation?
There is also RDA analysis in the papper (Werner et al. PNAS 2011), what is the major difference betweem RDA and CCA? When the assumption of the distribution of OTU is not unimodal, should we perform RDA?
RDA assumes linear response curves, in contrast to the symmetric, unimodal response curves of CA and CCA. RDA best preserves the Euclidean distance between samples while CCA best preserves the chi-square distance. as a result RDA is based on abundance and focuses on explaining differences in abundance between samples. CCA can best be thought of as explaining compositional differences between samples. CCA can also work fine for linear responses.
Nice video, Dan. Would it be interesting to consider the non-independence of chemotaxis and flagella (unless chemotaxis specifically does not include flagella genes in this case) and see how the results come out using an interaction (CHEMOTAXIS:FLAGELLA) in the CCA calculation?
There is also RDA analysis in the papper (Werner et al. PNAS 2011), what is the major difference betweem RDA and CCA? When the assumption of the distribution of OTU is not unimodal, should we perform RDA?
Good question. I have been looking for the same answer; however, I haven't found it :(
RDA assumes linear response curves, in contrast to the symmetric, unimodal response curves of CA and CCA. RDA best preserves the Euclidean distance between samples while CCA best preserves the chi-square distance. as a result RDA is based on abundance and focuses on explaining differences in abundance between samples. CCA can best be thought of as explaining compositional differences between samples. CCA can also work fine for linear responses.