LC-MS/MS normalisation techniques in metabolomics

Question

[judhenaosa]

Dear all,

I am working on metabolomics from LC-MS/MS data. This data was normalised using internal standards. As I know, some scaling approach like autoscaling or pareto scaling is enough to continue with the further analysis. However, the boxplots show in some cases an erratic distribution of the data across samples (Fig. 1). Even, when I run limma to identify differential abundance, the volcanoplot looks like asymmetrical (Fig. 2). In the last case, I am using PCA - limma to reduce the unknown technical source of variance.

My question are:

1. Is it possible to apply more sophisticated techniques for normalisation such as PQN, cubic, or quantile normalisation despite the data was corrected using internal standards?
2. Is PCA - limma a good choice for differential abundance analysis? Otherwise, Which approach do you recommend me?

Thanks in advance,

I am looking forward to hearing from you,

Juan