Identification of an unknown substance (e.g. a plant metabolite or an MS peak that your PCA analysis implies has predicitive value) can be tricky stuff. What steps would you say are required/enough to feel confident regarding the identity of a substance?
The Metabolomics Standards Initiative (MSI) is working on guidelines for all sorts of reporting, including the identiy of detected molecules.
See the paper and the website I have attached.
To quote from section 2.9 of the paper:
Currently, four levels of metabolite identifications can be found in the published metabolomics literature. They include:
- Identified compounds (see below)
- Putatively annotated compounds (e.g. without chemical reference standards, based upon physicochemical properties and/or spectral similarity with public/commercial spectral libraries).
- Putatively characterized compound classes (e.g. based upon characteristic physicochemical properties of a chemical class of compounds, or by spectral similarity to known compounds of a chemical class).
- Unknown compounds—although unidentified or unclassified these metabolites can still be differentiated and quantified based upon spectral data.
Authors should clearly differentiate and report the level of identification rigor for all metabolites reported.
The key section about positively identifying a substance is:
...it is generally believed that a single chemical shift, m/z value, or other singular chemical parameter is insufficient for non-novel metabolite identification. Thus, the following minimum standards for level 1, non-novel metabolite identification are proposed.
- A minimum of two independent and orthogonal data relative to an authentic compound analyzed under identical experimental conditions are proposed as necessary to validate non-novel metabolite identifications (e.g. retention time/index and mass spectrum, retention time and NMR spectrum, accurate mass and tandem MS, accurate mass and isotope pattern, full 1H and/or 13C NMR, 2-D NMR spectra). The use of literature values reported for authentic samples by other laboratories are generally believed insufficient to validate a confident and rigorous identification. The use of literature or external laboratory data result in level 2 identifications.
- If spectral (MS or NMR) matching is utilized in the identification process then the authentic spectra used for the spectral matching should be described appropriately or libraries made publicly available. It is preferred that the reference spectra are made available at no cost, but the CAWG recognizes that this may not always be possible for commercialized libraries (NIST, Wiley, etc.). However, the premise of this minimum is that authors document and provide the spectral evidence to validate the metabolite identifications. If the authors choose not to provide the experimental evidence to support the identifications, then the identifications should be reported as ‘putative identifications’.
- Metabolite identifications based upon additional orthogonal data (i.e. more than two) are highly advantageous, provide additional confidence, and are often necessary to provide unambiguous identification of stereo configuration. Additional data consistent with best chemical practices might include: selective solvent extraction, retention time, m/z, photodiode array spectra, λmax and εmax,chemical derivatization, isotope labeling, 2D NMR, IR spectra, etc.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3772505/
http://msi-workgroups.sourceforge.net/
I would say that the requirements differ from substance to substance.
Proteins: you need the sequence. The most common method today involve mass spec, de novo sequencing and/or database searches. It is has its own challenges but there are robust workflows available.
Alternatively, biochemistry methods such Western Blotting.
DNA: as above, only with different instrumentation
Peptides: very similar to the protein case. Easier, to assess identity, for the analyte is far smaller and less complex
Lipids: mass spec (and MS/MS to that) in combination with LC or GC can give you a lot of information. I would say that the fragmentation pattern is what can give you confidence in the identification.
Small molecules (e.g. TCA cycle intermediates): I would say that mass speck is, once again, the most common technique to obtain high confidence identification. Fragmentation studies are particularly powerful.
Pharmaceuticals, unknowns etc etc: I used a combination of mass spec and NMR to identify unknown in pharmaceutical products. I don't think there is anything new on this side... If we do not consider very specific cases.
All things said, what one need to define an unknown's identity? You need methods to prove its structure and its configuration (in the case of enantiomers, for example).
You can use whatever techniques, so far you can prove your results are correct and there is NO possibility of mis-identification.
The last step to get the most confident result would be comparison with a standard.
The Metabolomics Standards Initiative (MSI) is working on guidelines for all sorts of reporting, including the identiy of detected molecules.
See the paper and the website I have attached.
To quote from section 2.9 of the paper:
Currently, four levels of metabolite identifications can be found in the published metabolomics literature. They include:
- Identified compounds (see below)
- Putatively annotated compounds (e.g. without chemical reference standards, based upon physicochemical properties and/or spectral similarity with public/commercial spectral libraries).
- Putatively characterized compound classes (e.g. based upon characteristic physicochemical properties of a chemical class of compounds, or by spectral similarity to known compounds of a chemical class).
- Unknown compounds—although unidentified or unclassified these metabolites can still be differentiated and quantified based upon spectral data.
Authors should clearly differentiate and report the level of identification rigor for all metabolites reported.
The key section about positively identifying a substance is:
...it is generally believed that a single chemical shift, m/z value, or other singular chemical parameter is insufficient for non-novel metabolite identification. Thus, the following minimum standards for level 1, non-novel metabolite identification are proposed.
- A minimum of two independent and orthogonal data relative to an authentic compound analyzed under identical experimental conditions are proposed as necessary to validate non-novel metabolite identifications (e.g. retention time/index and mass spectrum, retention time and NMR spectrum, accurate mass and tandem MS, accurate mass and isotope pattern, full 1H and/or 13C NMR, 2-D NMR spectra). The use of literature values reported for authentic samples by other laboratories are generally believed insufficient to validate a confident and rigorous identification. The use of literature or external laboratory data result in level 2 identifications.
- If spectral (MS or NMR) matching is utilized in the identification process then the authentic spectra used for the spectral matching should be described appropriately or libraries made publicly available. It is preferred that the reference spectra are made available at no cost, but the CAWG recognizes that this may not always be possible for commercialized libraries (NIST, Wiley, etc.). However, the premise of this minimum is that authors document and provide the spectral evidence to validate the metabolite identifications. If the authors choose not to provide the experimental evidence to support the identifications, then the identifications should be reported as ‘putative identifications’.
- Metabolite identifications based upon additional orthogonal data (i.e. more than two) are highly advantageous, provide additional confidence, and are often necessary to provide unambiguous identification of stereo configuration. Additional data consistent with best chemical practices might include: selective solvent extraction, retention time, m/z, photodiode array spectra, λmax and εmax,chemical derivatization, isotope labeling, 2D NMR, IR spectra, etc.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3772505/
http://msi-workgroups.sourceforge.net/
Tandem mass with GC or LC shall increase the confidence level in MS peaks identified specially in case of known compounds. Combination of RT, parent ions and daughter ions can solve the issue. In case of unknown compounds, we have to generate multiple data. I think, the above two answers (Daniel and Fabrizio) explained everything. Thanks to both Experts (Daniel and Fabrizio) for such a nice and explanatory answers :)
Dear Steven, very good question !
Identification of an unknown substance is an inverse problem (in mathematics).
Direct problem: substance->properties, Inverse: properties->substance.
Theoretically there is no the exact solution since there are no absolutely pure substances in common chemical analysis (noise=mpurities). Maybe in atomic physics "yes" ?
We use a priory information about unknown substances (chemical data bases).
Identification of an unknown substance is done with some probability.
Hello. In case it is a metabolite, a mass spectrometry analysis and an 1H-NMR spectrum of the molecule matching the predicted structure are enough (+13C NMR if possible). Also can be added and especially if not possible to have one of the two mentioned above, an IR spectrum, an elementary analysis and/or X-Ray analysis.
It is difficult to add more after the excellent answers given by Fabrizio and Daniel but for small molecules I would recommend GC-MS (retention time, purity and separation, molecular mass+fragmentation pattern), IR (functional groups), 1H NMR (protons), 13C (carbon) and 2D NMR for sure. For known molecules the spectra are commonly placed in spectral databases
R- and S- stereoisomers can be problematic. Multiple positions of symmetry are worse. And R-, S-, R- with potential differences at each point of assymetry- difficulty!
Z vs E isomers about double bonds, the same.
Positional isomers will have different fragmentation patterns in GC-MS, but you have to know where to look.
As a reference it might be useful to look at the guidelines for detection of e.g. pesticides that are available in both the EU and the US (and other countries, for sure). Analytically these compounds are much more closely related to metabolites than e.g. peptides or proteins. The EU system works with identification point that are summed up - the more orthogonal data points (RT, accurate mass, MS/MS vs. reference db) you accumulate the more confident, and above a certain level you define the analyte as being identified. My expectation would be that the MSI that Daniel kindly pointed out will come up with a similar way of scoring ID data to define tangible thresholds.
For relatively small molecules, for instance a metabolite, an authentic sample provides the best verification. In some cases it may already be known while in others you may have to synthesize it. Whether or not it's worth the effort depends upon how important the exact identity is.
Check out this link (Frode Rise sent it to me), this is the research nightmare for mistaking identity of a molecule. I guess NOESY should have added to the mix of elucidation in this case:
"A promising anticancer agent about to enter human clinical trials is on the hook because of a chemical structure error discovered by scientists at Scripps Research Institute California. The patented compound, known as TIC10 or ONC201, is owned by the biotech firm Oncoceutics. However, Scripps has applied for a patent on the corrected structure and has licensed it exclusively to another company, Sorrento Therapeutics.
The reanalysis and subsequent licensing could lead to an unprecedented legal case—the first in which a structural reassignment puts in jeopardy a patent and clinical trials."
http://cen.acs.org/articles/92/i21/TugWar-Over-Promising-Cancer-Drug.html
When you use PCA you only see the distribution of differente kind of compounds in one or more extracts, to find a known compound you derreplicated using NMR or MS techniques, then you compare your data with different data bases and identify which data are different or unknown. A new compound has to be isolated and its stereochemistry has to be determined.
It is a loaded question. Answer will vary depending on the source of the compound. For a compound with a known structure, it will be the comparison of physical, chemical, chromatographic and spectroscopic data.
But for an unknown compound one has to perform a large battery of analysis starting with the analysis of the source followed by both physical, chemical, chromatographic and spectroscopic analyses and a lot of guess work and finally confirmation with synthesis of proposed compound.
our approach described at:
http://littlemsandsailing.files.wordpress.com/2013/12/lcgc_article.pdf
In LC-MS analysis, identification of a small molecule should depend on two criteria: firstly, the molecule detected in LC-MS spectra should have sufficient abundance. Secondly, it should have reproducible peak retention time on LC separation. Meanwhile, it is worth to note that such identified molecule may not the real molecule in your sample, but do exist in your analysis results. Such molecule may be artifacts or introduced by solvent. Therefore, we should conduct control experiment with blank sample to exclude such molecules. For more details, please refer to our recent paper.
Article Hydrophilic interaction chromatography coupled matrix assist...
Our appoach is described here:
Schymanski, E.L., Jeon, J., Gulde, R., Fenner, K., Ruff, M., Singer, H.P., Hollender, J. (2014): Identifying Small Molecules via High Resolution Mass Spectrometry: Communicating Confidence, Environmental Science and Technology, 48 (4), 2097-2098. DOI: 10.1021/es5002105
Dear Steven,
This is a very interesting topic on which I've been working on on and off for about 2 years, so I'll try to summarise it here for you.
Firstly, Daniel and Fabrizio touched upon this point very well. I personally work with mass spectrometry, so I will base my reply around that.
To prove beyond a reasonable doubt that you've identified a molecule you typically have to prove with multiple tests that you've made a positive identification. A few examples:
- using a high resolution instrument: confirmation by accurate mass and fragmentation pattern match with additional confirmation from isotopic pattern and/or retention time
- using multiple (low resolution) instruments: if you get a match for a GC-MS peak by doing a search in a library like the Wiley/NIST, then as additional confirmation you can try injecting the same compound on an LC instrument - if amenable(LC-MS / LC-UV) or a triple quadrupole instrument to get MS/MS information.
- as a general rule-of-thumb, buying the analytical standard and comparing with the peak / spectrum obtained in the sample is pretty much a golden standard in this
- although not as common as mass spectrometry instruments, NMR is a very valuable tool in providing more information about an unknown analyte, but it's not such a common instrument in analytical labs like MS instruments and you need your extract to be very clean and uncomplicated to make analysing the NMR data more easy
- you can even get some information about an analyte during sample prep: by doing a fractionation on SPE cartridges with solvents of increasing polarity (provides a polarity range for your analyte) or by switching the ionisation source on the same LC instrument and seeing if your analyte ionises with ESI, APCI or APPI.
Here is an interesting paper about communicating confidence in the identification of small molecules by HRMS: http://pubs.acs.org/doi/abs/10.1021/es5002105
And another article about the use of HRMS for this purpose, but also complementary techniques: http://www.ncbi.nlm.nih.gov/pubmed/21472594
From my experience I agree with the answers from Daniel and Alin. They are very complete and summarized. To identify an unknown compound you must use different instrumental analytic techniques MS (HRMS) (GC-MS or LC/MS it depends of the molecular mass of the molecule) NMR (H,C), FTIR. Each of them gives you a different information about the structure of the molecule. It is like a puzzle and with the different information you can build it.
Dear all,
Chemists need to develop clear methods of identification. Practical identification is intuitive.Mathematical methods of selection of sufficient information for identification purposes are rarely used. The usual approach is based on the principle: the more is the better.
Orthogonal identification data are required. MS spectrum identical to a reference compound is not enough. identical retention time in GC or LC may support the MS identification. If the molecule is radio-labeled and the cold reference available in sufficient quantities, crystallization to constant specific activity of the mixture may provide the proof of identity beetween the cold and the radio-labeled substance. Lipid properties are needed for such crystallizations. In the case of small molecules, chemically oxydized or reduced derivatives of both reference and unknown should be the same (have identical retention times at least).
Dear Robert Morfin,
Very interesting approach: Orthogonal identification data.
Mass spectroscopy associates with nuclear magnetic resonance are the main tools to identify molecules.
But for an unknown compound one has to perform a large battery of analysis starting with the analysis of the source followed by both physical, chemical, chromatographic and spectroscopic analyses and a lot of guess work and finally confirmation with synthesis of proposed compound.
---------------------------------------------------------------------------------------------------------------------
I fully agree with Dr. Akhtar.
- Badges
- Science method
- Similar topics
- Chemistry
- General Biochemistry
- Biochemical Techniques
- Analytical Chemistry