Matching issues

[JustinCarrard]

Hi, First of all, I would like to congratulate you on a great ontology tool, many thanks! I am experiencing a few matching issues while uploading our dataset (serum samples from healthy humans in their twenties and seventies, generated by using MS-DIAL) and I was wondering if you would have any advice on how to fix them.

The following SM are currently recognised as "ceramide phosphocholines (sphingomyelins) [SP0301]" only. How would you name them to let LION/web recognise them in more details ? I tried to convert the MS-DIAL nomenclature using the brand new LipidLynxX, but it wasn't really helpful here. SM 37:2;2O SM 39:1;3O SM d32:1 SM t41:1

The following FA is recognised as "fatty acids and conjugates [FA01]" only. I believe the reason being, LION/web considers it as exogenous. Is it correct ? Do you plan to add these species in the future ? FA 34:9

Same questions for the following TAG, which are identified as 3-4 different species by LION/web. TAG 10:0-12:0-14:0 (identified as 3 species: TG(36:0), C12:0 & C14:0) TAG 14:0-15:0-18:1 (identified as 4 species: TG(47:1), C14:0, C15:0 & C18:1)

These three lists are not exhaustive but made of chosen examples illustrating the problematic we have facing.

Many thanks for your help !

[martijnmolenaar]

Hey Justin,

Thanks for using LION/web!

All these difficult lipid dialects remain difficult for the field. I tried to covert the SMs that you mentioned with LipidLynxX, resulting in: SM 37:2;2O -> SM(37:2;2) SM 39:1;3O -> SM(39:1;3) SM d32:1 -> SM(32:1;2) SM t41:1 -> SM(41:1;3)

These results are matched to LION (although the ;2 and ;3-suffixes are neglected in LION on summed lipid level).

For, FA 34:9, do you refer to a fatty acid with 34 carbons and 9 unsaturations? This fatty acid is not present in LION (it is extremely long). The smart-matching option (switched on by default), does recognize the FA prefix as a fatty acid.

The same is true for the TGs. The species you mention are not present in LION. However, 'smart matching' does recognize the class (TG) and individual FAs. You can regard this as a dynamic version of the ontology, i.e. the analysis give the same results as TG(14:0/15:0/18:1) would be present.

Hope this helps!

[JustinCarrard]

Hi Martijn,

Many thanks for your detailed reply. I am new to the field of lipidomics (I was trained as a MD), so I really appreciate your tipps.

I could directly match most of the lipid species by now. I am just struggling to match the following 3 lipids: 1) LPE(O-16:1) 2) LPE(O-18:1) 3) PC(O-16:1/16:0)

LION/web recognises them as : 1) monoalkylglycerophosphoethanolamines and C16 2) monoalkylglycerophosphoethanolamines and C18 3) PC(O-32:1), C16:1 and C16:0

I reckon, they are not part of the LION/web library yet, are they ? I will rename the third one PC(O-32:1) to avoid LION/web recognising it as 3 different molecules. How could I rename 1) and 2) to avoid LION/web recognising them as two molecules, i.e. I just want to have monoalkylglycerophosphoethanolamines as a result ?

Yes, FA 34:9 is a extremely long FA (probably coming from the diet or microbiome), same for the TG I mentioned.

Many thanks for your support !

[martijnmolenaar]

Hi Justin,

This is a result of the setting 'smart matching' (in settings tab on the left). However, it is actually what you probably want. You can see the ontology is a hierarchical tree of nodes, with the individual lipid species at the bottom. PC(16:0/16:0) is linked to class PC, PC is phospholipid, etc.; it is also llinked to 16:0, 16:0 is a fatty acid with 0 unsaturations, etc. In case species are not present, the setting 'smart matching' will add these associations dynamically. For instance, PC(O-16:1/16:0) is now linked to PC(O-32:1), which is a alkyl-PC, phospholipid etc. It also dynamically links it to C16:1 and C16:0, including all the higher terms. Now, LION/web will behave as if PC(O-16:1/16:0) would have been present.

Attached a graphical representation of this.

So, I would recommend to not change these lipid names, and proceed with your analysis.

Best, Martijn

[JustinCarrard]

Hi again,

Ok thanks. I got the idea of the ontology but believed that it artificially added non-existing species in this particular case (C16:1). Now I perfectly understand it thanks :-) Best Justin

[JustinCarrard]

Hi Martijn,

I wanted to try the LION-PCA heatmap module but I am keeping experiencing the same error: "An error has occurred. Check your logs or contact the app author for clarification." See print screen below. I am using the exact same file as for the enrichment analysis, so format shouldn't be an issue. I retried the enrichment analysis and it does work perfectly. Do you have any idea how I could fix it ?

Also I wanted to give you a feedback regarding the enrichment analysis. I noticed that LION/web recognises Cer(d18:1/24:0) but not Cer(18:1;2/24:0) as Cer(d18:1/24:0) (direct match, LION:0001706). Conversely, HexCer(34:1;2) but not HexCer(d34:1) is recognised as HexCer(34:1) (direct match, LION:0000278). So with my dataset, I need to mix Lipidmaps and LipidLynxX notations to obtain the best outcome with LION/web. No criticism here, I really like a lot LION/web. I just thought, it could be of interest for you, therefore I checked the option "Automatically send unmatched lipids to LION-team".

Best Justin

[martijnmolenaar]

Hey Justin,

Hmm, interesting. It is hard to say what is going wrong. If you want, you can send me a mail (email address at https://www.uu.nl/staff/MRMolenaar) with your dataset so that I can have a closer look. Will delete the dataset afterwards.

Thanks for the other feedback, we will have a look at this.

Best, M

[vaishnavikaashyap]

Hi Martijn,

Firstly I would like to thank you as this tool is really helping me in my internship work related to enrichment analysis of Lipid-omics

Further, I have a similar query related to the Matching function of the tool.

As per the discussion above I understood that the tool divides the lipid and considers it based on class, carbons and unsaturated compounds. Having said that in the example of PC(O-16:0/16:1) it is mentioned that the PC(32:1) would be considered as an alternative (Which means it sums up the chain count).

Similar in my query I have CerP(d18:1/26:1) considered as - Ceramide 1-phosphates [SP0205] , C26:1 , d18:1 (sphingosine).

Why does the tool only show the class (as there will be multiple lipids under that class, which one it will consider to show in the network), why will not it sum the chains to CerP (d44:2) in this case ?

Could you please help me understand this ?

Reagards, Vaishnavi

[martijnmolenaar]

Dear Vaishnavi,

It should work as you suggest. I will check why it's not working and let you know.

Best, Martijn