Dear All
I am curious to know the actual procedure to prepare QC samples in metabolomics profiling. I would appreciate if you can shower your opinions on preparation of QC samples. Do we need to pool the biological sample for QC and extract the sample or can we pool the extracted samples from individual samples and prepare the QC. Your opinions or advise is much appreciated.
Regards
Pasha
What I did to prepare QC samples is by pooling extracted all individual samples that will be analyzed in one series of analysis. PCA analysis was applied to prove that our method is valid; for each series of analysis we should have own QC sample. See attach publication (although this publication is for urine, we can apply for other matrices)
Prof. Indrayanto
Thanks for the reply. Currently, i am going through the literature and i found that most of the papers support preparation of QC by pooling the biological matrix before extraction. So, i am curious to understand what the difference it makes in the data if we prepare the QC either by pooling before or after extraction.
Regards
Pasha
The purposes of QC samples is to prove that your method is valid (LC-MS), so it means that intra-day and inter-day precession is OK, by PCA this can be showed by all data of QC samples are closed each others (tight clustering) as shown in the references that I attached before. By QC samples you should prove that identical samples will give identical responses (in this case TIC intra-day or inter-day of QC sample.), So I think it is no need to do pooling before extraction (but if you want to do before extraction, it is also OK). But you should first check the stability of your samples by running the sample after keeping in certain time (o, 6, 12, 24 h etc), PCA should showed tight clustering of the data of samples in different time of incubation (0, 6,.....24h..etc). If you want to know the difference of the methods of extractions (different operators, different solvents, different equipments) than you should do before extraction.
Having a pool sample before extraction is the way forward; your samples should be treated the same way as your QC. This will ensure that all factors influencing your samples on different days or times also influence your QC.
I think it really depends what you are trying to do when you run a pooled biological QC, as to whether or not you pool before or after extraction. If you are trying to capture all the variation in your procedure then it may be relevant to pool before extraction, but then I believe you are mainly testing the reproducibility of your sample preparation technique - and really this ought to be corrected for by the addition of internal standards. On the other hand if you are using the pooled biological QC to correct for the behaviour of metabolites in your analytical system, that are not able to be corrected for by the internal standard (i.e. metabolites that behave in an opposite pattern to the internal standard), then you really should be pooling after extraction and then split into individual aliquots that are run throughout your batch sequence at regular interval. Under these circumstances you have created a homogenous identical mixture that theoretically should give you identical chromatograms, but in fact will reflect any variations in your analytical system. These variations can be corrected on a metabolite by metabolite basis. For a really good paper on this you should look at:
The importance of experimental design and QC samples in large-scale and MS-driven untargeted metabolomic studies of humans
WB Dunn, ID Wilson, AW Nicholls, D Broadhurst
Bioanalysis 4 (18), 2249-2264
David Broadhurst has done a lot of work around the development of algorithms to make use of these pooled QC's. Minimally though, as suggested above use the QC's as a way to assess the quality of your data when performing a PCA. After transformation and normalisation they should be a tightly clustered group that sits into the middle of your PCA plot.
The last answer by David is all you need to know really, the article by Dunn and coworkers is still valid and a very good read.
I would recommend that you, prior to running your analysis, use QC injections to prove that your method is fit for purpose and pre-condition system. Typically a combined use of PCA and univariate analysis is good. This requires quite a few injections, but most of the features observed (70% or more depending on who you ask) should have a %CV value below 30% (in both RT drift and intensity), be present in 2/3 (in theory all injections!) of the samples. If this is obtainable, I would consider running your real samples. There is still a need to run QC samples in between your real samples however (to prove system suitability across the batch). Its not fun to stick these in the batch without knowing that your method is fit for its purpose and then have to re-run everything again.
You can also check the article by Vorkas et al (2015) based on tissue, they highlight how they use QC samples quite well.
Article Untargeted UPLC-MS Profiling Pipeline to Expand Tissue Metab...
Sorry %CV for RT should be much lower of course, below 0.5% is acceptable I would say.
David P De Souza
Dear David, I haven't understand "... These variations can be corrected on a metabolite by metabolite basis." could you tell more ? Thanks
Best,
Xiangping Lin
The intention was to monitor the behaviour of individual metabolites throughout the sequence relative to instrument behaviour. For instance if a metabolite was seen to degrade between from one PBQC to the next, then the samples in between those PBQCs would be normalised to reflect that degradation. Since this original question was answered, we and others have found that this can lead to overcorrection, and in some situations loss of the biological variation you're interested in.
I would recommend that PBQCs are used, but more in the way that Mikael Karl Robert Engskog describes above. This is primarily the way in which we are using them now - to check analysis quality, and measure CVs of individual metabolites.@
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Control Theory
- Control Systems