I have come across different publications that utilise different methods for analysing the difference in microbial compositions when comparing different populations (using regression based analyses).
For example, some will work with just relative abundances that they have calculated from the operational taxonomic units (OTUs).
Other papers have normalised these relative abundances using, for example, log10 transformations.
Others have normalised the OTU counts, prior to utilising them in statistical tests (without calculating relative abundance).
Please could anybody elaborate as to which method is the better one to use?
Hi Anna,
Unfortunately, there is no definite answer--microbiome differential analysis methods are still in their infancy and the answer depends on your question. Some statisticians favor compositional data analysis (CoDa) methods because microbiota data, like other NGS 'omics data, is compositional in nature. Popular CoDa methods are ALDEx2 and ANCOM-II. Recently, these two methods were demonstrated to produce the most consistent results across many datasets, compared to other methods. See doi.org/10.1038/s41467-022-28401-w.
Best regards,
Ali
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publications