LFQBenchmark experiment - multiple organisms

[brvpuyve]

Hi everyone,

I generated an updated LFQbenchmark dataset, similar to the one from Navarro et al. (https://pubmed.ncbi.nlm.nih.gov/27701404/). I was wondering how I could best annotate the mixtures (as pooled samples)? Can I mention more than one organism in the characteristics[organism] column? Additionally, would it be beneficial to add an additional comment section to define the ratio's of the three proteomes?

Looking forward to your suggestions!

Best,

Bart Van Puyvelde

[mlocardpaulet]

Hi Bart, did you get any help with this? I suspect you could use the field characteristics[pooled sample] and list in it all the samples that are pooled (SN=sample 1,sample 2, …​ sample 9 were "sample n" is the value of the corresponding sample in source name). For the relative quantities I am not sure. Others may have better ideas. Maybe you could use the key QY= to indicate relative quantity (like in characteristics[spiked compound]), but I am not sure how to make the sample names correspond to the quantities.

[mlocardpaulet]

Also, I don't know how to do if one of the pooled samples is not analysed alone (so there is no .raw file associated to one of the sample names).

[ypriverol]

Hi @mlocardpaulet @brvpuyve :

First, my apologies for the late reply, I was OFF for a couple of weeks. I was discussing a some weeks ago about with @anjaf about how to represent multiplexed samples in an experiment.

We have two options here:

1- Represent each sample as an independent sample, adding a characteristics to the sample called characteristics[concentration of] and link each sample to the same data file. The characteristics[organism] will be different for each sample. This is actually a clean representation because each sample has its own row and can be represented with more characteristics. It has differences with the current pooled approach mentioned by @mlocardpaulet because in the pooled approach samples are used multiple times in their corresponding channel + in the pooled.

It will be something like:

source name	characteristics[organism]	characteristics[organism part]	characteristics[biological replicate]	characteristics[concentration of]	assay name	comment[technical replicate]	comment[fraction identifier]	comment[label]	comment[data file]	characteristics[concentration of]
Sample-1	homo sapiens	heart	1	70%	ms_run 1	1	1	label free sample	file1.raw	70%
Sample-2	e coli	liver	1	60%	ms_run 1	1	1	label free sample	file1.raw	60%

As you can see the assay name is the same meaning that the file and the label conditions are the same.

2- @anjaf mentioned before the idea of having an characteristics[organism] called mixed, then we can represent all the species in the sample in the characteristics[pooled sample] as key values pairs with concentrations.

Would be great to have your opinion @anjaf @jgriss @mvaudel @mlocardpaulet @all @bigbio/collaborators

[mlocardpaulet]

Hi @ypriverol thanks a lot. I like option 1- very much. So to be clear: there will be duplicated file names?

[brvpuyve]

Option 1 is maybe the best approach although it will be some work for me to add the extra lines :-) Let me know what is decided and I will create the SDRF's.

Thanks for the comments!

[ypriverol]

Hi @ypriverol thanks a lot. I like option 1- very much. So to be clear: there will be duplicated file names?

Yes. We have the same case when multiple samples are multiplexed in the same RAW file.

[enryH]

I guess option one is fine if the python client can identify such a case?

• each raw file, if not unique is a mixture?
• should concentration add to 100%? (to be valid?)

[jgriss]

Hi all,

We already have this case covered in some sorts for isobarically labelled experiments (see PXD017799 as an example). Here, we also have mixtures of multiple, independent samples in one raw file.

I therefore strongly suggest to stay consistent with the design approach that was chosen there, which essentially is what @ypriverol mentioned as option 1.

In case of isobarically labelled experiment, this could even be extended to have multiple rows referencing the same channel in the raw file.

@enryH

• Personally, I think that characteristics[concentration of] should be optional, but if provided must add up to 100% to be valid
• In isobarically labelled experiments we also refer to each raw file multiple times indicating that it's a mixture. But we might not always have / need f.e. the individual sample concentrations - just to keep this case in mind as well

[mlocardpaulet]

Hello again, sorry it took me so long to come back to this. I am looking at the headers that have been utilised in the SDRF generated to date and I see that characteristics[concentration of] is used to define the concentration of compounds defined in characteristics[compound]. So if we go with the option 1 (if I understood well: one row per sample in the pool, with the respective quantities annotated in characteristics[concentration of]), can you distinguish the 2 usages of characteristics[concentration of]? Could this be an issue?

[enryH]

Hmm. If there is characteristics[organism] and characteristics[compound] then I guess it has to be ordered, but I am not 100% sure about this:

characteristics[organism]	characteristics[concentration of]	characteristics[compound]	characteristics[concentration of]

Could you explain the type of experiment where this is an issue?

But I agree that this could be an issue if it leads to ambiguous interpretations.

[mlocardpaulet]

Hello,

I guess you are right, I cannot see an example where it would be used.