I think to do metabolite profiling, you must do with validation first of your method, you should do repeatability studies (intra-day, same equipment, same column, same analyst) and do intermediate precision (at least 3 different days,  different analyst, could be different equipments/column but should in the same laboratory).

If your method is valid than you can do with different equipments and column etc.  You can use (+) or (-) ion modes or both, after these you can evaluate your data using multivariate analysis (PCA, PLS, HCA etc).

For detail please read; J-L. Wolfender et al., J. Chromatogr. A  1382 (2015)  136-164; E. J. Want et al., Nature Protocol, 5 . No. 6 (2010),, 1005-1017.

The relevance of using the same machine and the same column for measurements in positive or negative mode depends on what you want to achieve. It is not important if you exclusively want to use e.g. a PCA model to predict if an unknown sample belongs to a particular class, without reflecting on the analytical signals. As soon, as you want to try identifying or characterizing a relevant metabolite by its analytical signals, it becomes more important.

In any case, the retention times within each individual series of measurements should be perfectly comparable. Unless there is a particular reason to use distinct chromatographic systems I would suggest to have all samples (positive and negative mode) run on the same machine with the same column and even in one sequence.

see Want et al. (2010) Global metabolic profiling procedures for urine using UPLC–MS. Nature Protocols 5: 1005.

Hi:

1) To integrate the data it exists different algorithms. In our company, we use, mainly, mzMine open source platform ( http://mzmine.sourceforge.net/). 

2) The same mzMine software can be used to allign the extraction mass chromatogram peaks obtained by the different analysis. When they are alligned and normalized, it is possible to calculate the relative area ratio that is generally correlated with the amount metabolites ratio. It is difficult to correlate negative and positive ions because not all molecules ionize in both modality. I suggest to obtain two different mzMine data extraction table: a) one for positive and b) one for negative ions. After statistical analysis, it should be possible possible to identify the compounds and to search some metabolism correlations using the open metabolite database (Kegg, Metlin etc).

3) It is important to work in stable, standardized and reproducible conditions so to reduce the problem occouring during the allignment phase (see point 2).

I hope that these suggestions are of help for you. Do not exitate to contatct me to obtain further informations about this topic.

Simone

Stumbled into the same question. I suggest checking this software MSCombine to merge positive and negative modes.

Article MSCombine: A tool for merging untargeted metabolomic data fr...