Hello！I used the same data set to fit an opls-da model in metaboanalyst and simca respectively.Differential metabolites were then screened by the metric of VIP values greater than 1 under this model.By comparison, it was found that the differential metabolites screened using metaboanalyst and simca were very different and their VIP values varied considerably.I’d like to ask why that is. I would appreciate it if could give me some advice.
MetaboAnalyst is not just a statistical program. It contains comprehensive data processing and normalization steps to make sure the input are appropriate for statistical analysis. All analysis procedures are based on well-established practices in metabolomics. You can trace all the underlying R commands together with MetaboAnalystR package.
It is hard to compare with a commercial tool as we have no knowledge about its underlying code. I suspect it could be related to data processing, normalization and VIP computing. Note the underlying R package is ropls
Thank you for your help! I used metaboanalyst to process the data, normalization it, and then import simca for model fitting, which kept the previous data preprocessing steps consistent, but there was still a big difference in metabolites screened by vip value. Isn’t the principle of vip value calculation the same? So could this difference be due to coding?