After reading this paper:
http://journal.frontiersin.org/article/10.3389/fgene.2016.00191/full
I would like to know what are the pros/cons of using non-metric multidimensional scaling for RNAseq or other enrichment experiments:
https://en.wikipedia.org/wiki/Multidimensional_scaling#Non-metric_multidimensional_scaling
In comparison with other DGE methods, it appears that the primary difference between DGE and NMDS is that DGE analysis is a pairwise comparison, whereas NMDS is classed as an ordinate analysis:
where all sites for each sample are integrated as one pattern and the patterns are then compared across samples to identify sites that are conserved within a sample group while also divergent between different sample groups
The paper mentions the R package vegan as containing several robust ordinate analyses.
It's the first time that I come across NMDS in this type of datasets and I would like to know people's impressions on it. Anyone?
0 answers
No answers yet.
Log in to answer this question.
DGE is not a pairwise comparison.
At the end of the day whether NMDS is useful for you will depend on what sort of question you want to ask. If you care about genes that change then it's not useful. If you're interested more in finding large classes of differences where individual class members are not necessarily biologically relevant then NMDS can be useful.
NMDS is used quite a lot in 16S profiling to differentiate samples based on their profile. It can be used to differentiate samples based on their RNA expression levels. Look for phyloseq for examples.