What does the "Ions" parameter do in the Detect Compounds node?

[michaelseelion]

I am using Compound Discoverer to find what compounds are present in my untargeted metabolomics study. I hope someone can explain to me what does the Ions parameter in the Detect Compounds node do in Compound Discoverer 3.3. More specifically, I read the help document and learned from training events that you should choose a list of possible adducts present in your data for this parameter, but I do not understand how this list is used in the analysis pipeline. Say I have a peak at m/z=100.0314 and rt = 12.01 min (made-up numbers), and I choose [M+K]+, [M+H]+, and [M+Na]+ for the Ions parameter, how can the software tell which adduct it is and what compound it is an adduct of? 

[mnewmeyer]

Hi, @michaelseelion! I'm happy to share some of my experience working with CD to help answer your question! Using your example, if a feature with m/z = 100.0314 (assuming this is your [M+H]+ feature) and RT = 12.01 min is detected, and a feature with m/z = 122.0133 is also detected (corresponding to [M+Na]+) at the same Rt (+/- a tolerance), and you have [M+Na]+ selected in the Ions parameter, then those two features would be considered as coming from the same compound and be grouped.

However, as you mention, how does the software "know" this is correct? I've run into issues when I select [M+H]+ and [M+NH4]+ as my Ions, and I sometimes see two rows in my Compounds results table: one row where the two features are grouped together and considered as coming from the same compound, and another row where the higher mass feature (previously the [M+NH4]+) is actually assigned the [M+H]+ adduct.

The short answer is: it doesn't know which is "correct". It's taking your input parameters and seeing if the detected features fit that isotopic pattern and, if so, groups them together as adducts of the same compound. It's up to the user to go through and evaluate those results and see if they make sense. For most compounds, the [M+K] and [M+Na] adducts are going to be much less abundant than the [M+H]+. A trickier scenario is [M+NH4]+ because there are compounds for which this is the dominant adduct. If you're running a mobile phase system without any ammonium additives, then you don't have to worry about that.

Hope this was helpful. Finding the right set of parameters for your analysis can be a tricky process and may require some iteration before you settle on the settings that suit you best. I've found the best way to learn the software is to analyze a mixture of known compounds through your method, and try processing the raw data with different workflows and compare the results.

[michaelseelion]

Thank you for your explanation! Let's see if I am understanding it correctly. So continuing the example, the software assumes all compounds at least form a [M+H]+ adduct. Then starting at a peak, assuming it was from a [M+H]+ adduct, the software looks for peaks with m/z values corresponding to the [M+Na]+ and [M+K]+ adducts within a retention time window. If those peaks were found, their isotopic patterns would be compared. If they were from the same compound, they should have similar patterns, and this is how they are grouped.

In our case we do add some ammonium acetate and ammonium hydroxide to the mobile phase for analysis of polar metabolites. So if the software treats a dominant [M+NH4]+ adduct as a [M+H]+ adduct, it not only won't find other adducts correctly, it will also get the wrong exact mass to search the databases with. Will adding [M+NH4]+ to the Base Ions parameter in our workflow help?

Another issue I can see is with low abundance compounds. Their +H, +K, +Na adducts are unlikely to have matching isotopic patterns and will be treated as [M+H]+ adducts of different compounds, but I don't think this is something that can be fixed on the software side.

[mnewmeyer · Accepted solution]

That sounds mostly right. I think the isotopic pattern is only applied to the A0 peak (the [M+H]+ adduct in your example); the +Na, +K, etc. adducts are based on the mass difference from the [M+H]+. In other words, I don't think you'd see a feature annotated as a +Na adduct of an isotope.

I neglected to mention the effect of the Base Ion parameter in my post above. The issue I had, I think, was because I did include the [M+NH4]+ adduct as a possible base ion. The base ion parameter tells the software which adducts are allowed to be the base ion in an isotopic pattern (i.e., the most abundant). As you mentioned, if you have a compound that truly forms a more intense [M+NH4]+ adduct than a [M+H]+ adduct (e.g., like long-chain PEGs) but you don't include the [M+NH4]+ as a possible Base Ion, then the feature will likely be assigned [M+H]+, depending on your settings, and that would affect the subsequent database searches (since the searches are based on the back-calculated neutral mass). However, including both the +H and +NH4 adducts as Base Ions may also have other unintended consequences, like the issue I mentioned above. Your decision about what to include as a Base Ion will depend, in part, on what kind of analysis you're doing (e.g., non-targeted vs suspect screening) and if you know anything about the contents of your sample, and your tolerance for false annotations and manual review.