Libra on MS3-based TMT quantification

[Shagun Gupta]

Hello

Our lab has been trying to use Libra to quantify MS2 (Ion-trap) identification and MS3-based quantification of TMT datasets and we were trying to benchmark with an artificial yeast and mammalian spiked dataset with known fold-change (FC) values. However we observe drastically different values than expected, something we don't observe with other search engines for the same dataset. 

Has this issue been encountered before/ is there something obviously wrong when running with TPP that might cause this? Happy to provide further details on the dataset and parameters used to run with (most of them default apart from additions like static modification for TMT and specification of MS3 for quantification among others).

Thank you,

Shagun

[David Shteynberg]

Hello Shagun,

Thank you for your email and interest in the TPP.  I have recently been comparing TMTPro analysis by ProteomeDiscoverer and Byonic to that produced by TPP's Libra.  As far as I can tell, when I run and compare the quantities (intensities) they are mostly the same between Libra (without isotopic impurity correction and 0 pseudocounts) and the ProteomeDiscoverer/Byonic.  Execution of Libra is defined in the conditions.xml file that has to be defined for each Libra run.  I would be happy to take a look at your data and analysis to see if it can be placed on the right path for the Libra analysis to work.  Please post your results somewhere I can download and test.

Cheers,

-David

--
You received this message because you are subscribed to the Google Groups "spctools-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spctools-discu...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/ce2c75fb-2da1-4e05-8d08-bbe5a30b0495n%40googlegroups.com.

[Shagun Gupta]

Hi David

Could you suggest a good email to reach you with? I can share the pep.xml's and Libra condition file that way?

-Shagun

[David Shteynberg]

Please place your files on a shared drive and send me a link.

To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/7a0b1188-7c5f-4528-9e90-282b860e7627n%40googlegroups.com.

[David Shteynberg]

Hello Shagun,

Also please include your mzML files, which contain the actual data that Libra quantifies.

Thanks!

-David

[David Shteynberg]

Hello Shagun,

I noticed in your condition file you are using TMT10 with the following masses:

 <reagent mz="126.127726" />
    <reagent mz="127.124761" />
    <reagent mz="127.131081" />
    <reagent mz="128.128116" />
    <reagent mz="128.134436" />
    <reagent mz="129.131471" />
    <reagent mz="129.137790" />
    <reagent mz="130.134825" />
    <reagent mz="130.141145" />
    <reagent mz="131.138180" />

The difference between neighboring channels is <0.01 at the lowest and yet you are using tolerance of 0.2:

<massTolerance value="0.2" />

I think the appropriate mass tolerance for this type of labeling should be ~0.001.

Does that make sense?  Please try running Libra with the mass tolerance appropriate for this type of label. 

Cheers,

-David

To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/7a0b1188-7c5f-4528-9e90-282b860e7627n%40googlegroups.com.

[Shagun Gupta]

Hi David

Will do so, thank you! That makes a lot of sense. I have also added the mzXML files but this might be the cause of the discrepancy I see!

Thanks

Shagun

[David Shteynberg]

Hello Shagun,

Perhaps I am misunderstand here.  As far as I know, TMT labels are isobaric and are quantified in the fragment ion spectra.  Any peptide quantified would have to have the same precursor mass across all TMT labels.   Presumably there are not that many peptides that are identical between yeast and human.  The difference in your ratios will only be observable in the conserved peptides between human and yeast.  Do you have examples of specific spectra we can consider that work with this approach?  I was able to run the TPP tools including Libra on this data.

Cheers,

David

On Sat, Mar 26, 2022, 8:11 PM Shagun Gupta <sg2...@cornell.edu> wrote:

Hi David

So I reran with the changed parameter and the issue still seems to persist. I can share the updated results if you'd like as well? Also just confirming libra1 - 126, libra2 = 127N and so forth for a TMT10plex for example, since the issue is much aggravated in 2 out of three possible comparisons. To explain the setup more, yeast proteins are spiked with mammalian proteins such that yeast proteins are 10:4:1 (three replicates each) with mammalian proteins being (1:1:1) for the same.

Shagun

To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/ff3c5116-ca46-4aec-9ee4-d07d574db90an%40googlegroups.com.

[Shagun Gupta]

The zip folder for the files can be found here: https://drive.google.com/drive/folders/1hnR2igUXM_K-Pld2xspikeu_xnLGUdMc?usp=sharing

Let me know if you have any issues.

Best

Shagun

[Shagun Gupta]

Hi David

I've added the files on this drive link:

https://drive.google.com/drive/folders/1hnR2igUXM_K-Pld2xspikeu_xnLGUdMc?usp=sharing

Let me know if you have any issues accessing the files!

-Shagun

On Friday, March 25, 2022 at 1:34:57 PM UTC-4 David Shteynberg wrote:

[David Shteynberg]

Hello Shagun,

Thank you for sharing your complete dataset.  I still found the following in the conditions file you included:

TMT10_VB297695_condition.xml

  <massTolerance value="0.2" />

This should be set to 0.001 for this analysis to work on your type of label.  I am attaching the condition file with this change included so you can rerun Libra using the condition file attached.  Please make sure the tolerance is correct in the file when you try running Libra again and if you still see a problem please provide your TPP results interact* files etc...

Cheers,

-David

To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/4d2752de-2e7d-46fe-be38-5700043ca984n%40googlegroups.com.

[Shagun Gupta]

Hi David

Apologies, for some reason my post that it all worked out never made it through. Instead one of my older replies I think got posted again. Sorry for the confusion and thank you for your help!

Shagun

[David Shteynberg]

Great!  Thank you.

To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/43abc423-6593-469a-9ca1-b1edbd7c4c58n%40googlegroups.com.