We use MestreNova, but we don't really do metabolomics by NMR, so it may not suffice for your purposes.

AMIX and CHENOMX are two different softwares, they offer various performances.

With AMIX one can collect and organize 1D NMR and 2D NMR spectra (not only, even MS spectra)  in project folders and then visualize them, assign signals MANUALLY with help of  the contemporary inspection of 2D experiments  (ie TOCSY and HSQC) and eventually integrate them (1D and 2D spectra) with standard methods. Moreover, one can access to the statistycal tools in order to perform PCA and PLS (not all sofisticated routines, maybe OPLS is not implementated). One can easily process 1D spectra  with 'bucketing' selection (it offers many options for size, integration mode, exclusions ..) and obtain the variable X matrix (each variable is a section of your spectrum). At this point,  you can run the 'inner' mutivariate data analysis routines (PCA, PLS or PLS-DA) or you can export the data matrix so to process it with other softwares as SIMCA or on-line routines (metaboanalyst, R or else).

CHENOMX is a very good sowftare to perform quantification on 1D NMR spectra. I mean, absolute quantification with a great signal deconvolution routine. It allows one to perfectly match lineshapes of your spectral profile with a fitting procedure reproducing metabolites lineshapes according to a very extended library.  It is very easy to work with, and you will be able to fit the most part of your spectrum with a considerable number of metabolites,  automatically supplied with their concentration (with respect to your internal standard of known concentration..such as TSP). At the end, one obtains an X data matrix with  absolute concentrations for each metabolite, rather than ambiguous and over-estimated  integrals associated with buckets to be manually assigned. The proposed algorithm is really important to avoid over-estimation. So, the real benefits are: a reduced data matrix (just concentrations, so not too many spurious variables), absolute concentrations which can immediately reflect the biological conditions you are investigating. Moreover, signal identification is extremely accurate because there is a large and accurate library you can resort to, which is  for different solvents, pH values  and magnetic field strenght. Chenomx does not allow 2D spectra visualization and  any kind of math or statistical analysis.

In my opinion, neithe AMIX, nor CHENOMX  offers good spectra processing routines such as phase correction, baseline correction and so on. In my opion, the best in this case is MestreNOVA.

Sounds to me like you need MestreNova on the front end and one of the other packages on the back end! Are you doing human metabolomics or plant metabolomics?

John, we already have MestreNova. And we are going to process NMRs from blood and urine samples, of mouse, rats and human origin.

MestreNova is indeed a great package for doing structure elucidation and managing 1D, 2D and even MS data. We have used it for doing chemical fingerprinting of microbial and plant extracts, and perform multivariate analysis. However, we need something more robust to manage data and compare it with databases. 

Debora's description of both softwares is very helpful and I agree, that MestreNova provides the best in processing spectra.  I even prefer Mestre than topspin for processing my data.

In the end, however, when comparing AMIX and Chenomx, it seems that in one end you have AMIX that has more or less the same features as MestreNova, but with the addition of MS (Mestre is not yet ready for doing chemical fingerprinting analysis of MS data, just analysis of pure samples) versus Chenomx, which has the versatility of a powerful pH data-dependent analysis+concentration tools.

Sounds like Chenomx, then!

Giselle,

if you need a fast, cheap and trustworthy way to get your serum/plasma and urine spectra profiled I recommend you the webportal http://bayesil.ca/ from Wishart's group.

Just be aware that in order to get good profiling performances you should follow their protocols both in terms of sample preparation and NMR data collection.

Indeed Chenomx is a very good software for 1D NMR profiling of biofluids. And if you have to deal with tens or even hundreds of spectra probably it is the right solution for you.

I think you should also consider MNova for mixture analysis. We have recently released SMA http://mestrelab.com/software/mnova/sma/ which is now coupled to database plugin and it will be also coupled to PCA tool within few months.

In my experience having the possibility to work with both 1D and 2D and Jres at the same time is very helpful during compound identification; plus global spectral deconvolution could really make the difference when dealing with huge protein background that mask aliphatic region and make quantification almost impossible.

http://bayesil.ca/

http://mestrelab.com/software/mnova/sma/

Thanks for both tips, Silvia. I did not know Bayesil. 

In terms of database, I think Chenomx is currently the best choice out there, particularly if you are going to work with biofluids of a human origin. However, it does lack features for structural elucidation of compounds, so may not be a good choice if you have a lot of unknown metabolites (e.g. plant/microbial metabolomics).

While Bayesil is a nice program, my personal experience is that the automatic phasing and baseline correction still leaves a lot to be desired. I've now gone back to phasing and correcting the baseline manually for my spectra. Of course, I am only dealing with ~20 samples at a time (our machine time is pretty tight); if you are doing it for thousands of samples, then Bayesil is a good choice that will make your life easier.

I concur with Silvia that incorporation of 2D J-resolved experiments makes chemical identification much easier. However, I could never get used to the interface under Mnova (and this from a person that started from the clunky interface of XwinNMR). My personal set-up is Chenomx + iNMR, but the bookkeeping does become a problem. I've been searching for integrated solutions, but it seems that currently the landscape on the software side is still quite fragmented.