Questions regarding the philosopher 4.0.0 and MSFragger 3.3

[quark-spec]

Dear developers,

Thank you for the the latest release of philosopher and MSFragger. I tried it with the philosopher pipeline option, and out of curiosity compared the results to the old results obtained with philosopher 3.4.13 with MSFragger 3.1.1. A few questions arose from there

1) The number of PSM, peptide and protein all decreased in the latest version. This happened for both total and phospho-enriched proteome data. May I know if this is normal and why this is the case? 2) My data are TMT-labeled, so TMT-integrator was also run as part of the philosopher pipeline. However, the TMT-integrator reports are all empty for the phospho-enriched data, but not the total proteome data. The newest TMT-integrator 3.0.0 was used, so version shouldn't be an issue? 3) When I looked into the PSM.tsv and protein.tsv, there are a few proteins that showed up in protein.tsv, but are absent from the PSM.tsv. This also happened for both total and phospho-enriched proteome data. I felt this odd, as protein results are built on PSM information. May I know what may be causing this discrepancy?

Sorry for so many questions. I like the tools you developed very much. I just wish to learn more about the latest release. Thank you for your help.

[anesvi]

Something that you and Felipe would need to investigate, but a few extra questions:

Was if a single TMT-plex? Or multiple? If multiple, protein.tsv may have proteins identified from other plexes ( with 0 PSM in that plex)

Empty TMT-I file for phospho - did you run ptm-prophet? Also you see localization info in PSM.tsv?

Anyways, we tested extensively with FragPipe but Felipe need to see why philosopher command had issues

Best Alexey

Sent from my iPhone

On Aug 4, 2021, at 11:22 PM, quark-spec @.***> wrote:

 External Email - Use Caution

Dear developers,

Thank you for the the latest release of philosopher and MSFragger. I tried it with the philosopher pipeline option, and out of curiosity compared the results to the old results obtained with philosopher 3.4.13 with MSFragger 3.1.1. A few questions arose from there

1. The number of PSM, peptide and protein all decreased in the latest version. This happened for both total and phospho-enriched proteome data. May I know if this is normal and why this is the case?
2. My data are TMT-labeled, so TMT-integrator was also run as part of the philosopher pipeline. However, the TMT-integrator reports are all empty for the phospho-enriched data, but not the total proteome data. The newest TMT-integrator 3.0.0 was used, so version shouldn't be an issue?
3. When I looked into the PSM.tsv and protein.tsv, there are a few proteins that showed up in protein.tsv, but are absent from the PSM.tsv. This also happened for both total and phospho-enriched proteome data. I felt this odd, as protein results are built on PSM information. May I know what may be causing this discrepancy?

Sorry for so many questions. I like the tools you developed very much. I just wish to learn more about the latest release. Thank you for your help.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/Nesvilab/philosopher/issues/259, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AIIMM6Z5ECI57ORNXG4HT2DT3H7XHANCNFSM5BSRIL7Q. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email.

---

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

[prvst]

Hi @quark-spec. We are going to need additional information from you, more specifically the numbers you are getting. As Alexey mentioned, we performed a lot of testing, and we did not see any noticeable drop in numbers. Please send us a description of what kind of data, and experiment you have with the filtering log. We also need the version of all programs you used (i.e., MSFragger, Philosopher, and TMT-I), and all the parameters from both runs.

[anesvi]

Also, I only tested FragPipe. Can you run via FagPipe and see if you get the same problem as when using philosopher pipeline command?

Sent from my iPhone

On Aug 5, 2021, at 11:27 AM, Felipe da Veiga Leprevost @.***> wrote:

 External Email - Use Caution

Hi @quark-spechttps://github.com/quark-spec. We are going to need additional information from you, more specifically the numbers you are getting. As Alexey mentioned, we performed a lot of testing, and we did not see any noticeable drop in numbers. Please send us a description of what kind of data, and experiment you have with the filtering log. We also need the version of all programs you used (i.e., MSFragger, Philosopher, and TMT-I), and all the parameters from both runs.

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/Nesvilab/philosopher/issues/259#issuecomment-893550451, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AIIMM6YLPEAAGJKB62OJDZDT3KUT7ANCNFSM5BSRIL7Q. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email.

---

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

[quark-spec]

Hi @prvst and Alexey. Thank you for your reply. The experiment has 2 TMT plexes, with 11 channels in each, and data were generated from Orbitrap Lumos. Here are the filtering logs and parameter files from the runs using the old version of philosopher/MSFragger and also the new version. Files_for_philosopher_lastest_release_issue.zip

For the versions, the old results were from MSFragger 3.1.1+philosopher 3.4.13+TMT-integrator 2.1.5, and the new results are from MSFragger 3.3+philosopher 4.0.0+TMT-integrator 3.0.0.

I will also run via FragPipe and see if the problem is still there. Thank you again.

[quark-spec]

Sorry I missed Alexey's first comment.

Yes the experiment has 2 TMT plexes, but protein.tsv files are located under the directory of each TMT plex, so they should be specific to that plex?

And yes I did run PTM-Prophet, and I did see the columns like M:15.9949, M:15.9949 Best Localization, STY:79.9663 and STY:79.9663 Best Localization. I presume these are localisation information.

Thanks again.

[quark-spec]

I ran FragPipe 16.0. TMT-integrator output is still empty. However, the numbers of PSM, peptide and protein are closer to the old result and the drop in these numbers is not as significant. There is also not protein that only show up in protein.tsv but not PSM.tsv. Here are the FragPipe configuration files. Thanks again

Files_FragPipe16.0.zip

[anesvi]

Protein.tsv in each folder is “semi-specific” to each plex. It shows all proteins passing global FDR and PSM spectral counts from each specific plex. Others noted this inconsistency too, we should fix in philosopher ( at least do not print proteins with 0 PSMs in that experiment)

Need to investigate why the file empty.

Sent from my iPhone

On Aug 6, 2021, at 12:19 AM, quark-spec @.***> wrote:

the experiment has 2 TMT plexes, but protein.tsv files are located under the directory of each TMT plex, so they should be specific to that plex?

---

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

[quark-spec]

Sorry for asking but is there any update on figuring out why the noticeable drop in numbers only occurred in philosopher pipeline and the TMT-I files are empty.

[prvst]

Can you confirm to me that you are running a Phospho analysis with TMT plexes containing a reference channel called Pool?

[quark-spec]

Yes I did

[quark-spec]

And same for the total proteome TMT analysis, which had the same problems of large drop in numbers and empty TMT-I files.

[prvst]

I'm sorry, but based on your log I cannot find any issues with your run.

[quark-spec]

The drop in numbers happen in command line philosopher but not in FragPipe. Does that mean there is somehow an issue with command line philosopher?

[anesvi]

It must be. Perhaps you should schedule a zoom call with Felipe and Fengchao, so Felipe can better understand what parameter in the philosopher pipeline causes a drop in the IDs

From: quark-spec @.> Sent: Tuesday, August 24, 2021 11:16 PM To: Nesvilab/philosopher @.> Cc: Nesvizhskii, Alexey @.>; Comment @.> Subject: Re: [Nesvilab/philosopher] Questions regarding the philosopher 4.0.0 and MSFragger 3.3 (#259)

External Email - Use Caution

The drop in numbers happen in command line philosopher but not in FragPipe. Does that mean there is somehow an issue with command line philosopher?

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/Nesvilab/philosopher/issues/259#issuecomment-905148745, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AIIMM6Z65WHKTEPN3IYT5UTT6ROARANCNFSM5BSRIL7Q. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email.

---

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

[prvst]

I don't think there's something wrong, it's the same program, and the same functions, there's nothing specific to one or another. You are probably running different parameters with each program, can you list the ones you used with FragPipe, and the ones you used with the pipeline

[anesvi]

Yes, I mean so we can adjust some defaults in the philosopher pipeline to match what we have in FragPipe

From: Felipe da Veiga Leprevost @.> Sent: Wednesday, August 25, 2021 10:09 AM To: Nesvilab/philosopher @.> Cc: Nesvizhskii, Alexey @.>; Comment @.> Subject: Re: [Nesvilab/philosopher] Questions regarding the philosopher 4.0.0 and MSFragger 3.3 (#259)

External Email - Use Caution

I don't think there's something wrong, it's the same program, and the same functions, there's nothing specific to one or another. You are probably running different parameters with each program, can you list the ones you used with FragPipe, and the ones you used with the pipeline

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/Nesvilab/philosopher/issues/259#issuecomment-905533463, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AIIMM65CZMHHQPS5VAKZK7LT6T2N7ANCNFSM5BSRIL7Q. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email.

---

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

[fcyu]

I think I found something.

1. Philosopher pipeline runs PeptideProphet with all pepXML files from the same experiment together, while FragPipe runs PeptideProphet for each pepXML files separately (in closed search).
2. Philosopher pipeline uses max_variable_mods_combinations = 5000, while FragPipe uses max_variable_mods_combinations = 10000 for phospho analysis. But I am not sure if this difference matters much. Can set it to 10000 in Philosopher to see the difference.
3. After parameter optimization, the old MSFragger got 7 ppm fragment tolerance, while the new MSFragger got 10 ppm fragment tolerance. To figure out if this difference matters much, can change calibrate_mass to 1 to disable parameter optimization, and set fragment tolerance to 10 ppm.

Best,

Fengchao

[anesvi]

Should also ask if the drop happens in PSM.tsv etc. files from philosopher, or in TMT-Integrator files (if TMT data). To see if it I MSFragger/PeptideProphet related or something else. Make sure to run the same version of MSFragger too

From: Fengchao @.> Sent: Wednesday, August 25, 2021 10:34 AM To: Nesvilab/philosopher @.> Cc: Nesvizhskii, Alexey @.>; Comment @.> Subject: Re: [Nesvilab/philosopher] Questions regarding the philosopher 4.0.0 and MSFragger 3.3 (#259)

External Email - Use Caution

I think I found something.

1. Philosopher pipeline runs PeptideProphet with all pepXML files from the same experiment together, while FragPipe runs PeptideProphet for each pepXML files separately (in closed search).
2. Philosopher pipeline uses max_variable_mods_combinations = 5000, while FragPipe uses max_variable_mods_combinations = 10000 for phospho analysis. But I am not sure if this difference matters much. Can set it to 10000 in Philosopher to see the difference.
3. After parameter optimization, the old MSFragger got 7 ppm fragment tolerance, while the new MSFragger got 10 ppm fragment tolerance. To figure out if this difference matters much, can change calibrate_mass to 1 to disable parameter optimization, and set fragment tolerance to 10 ppm.

Best,

Fengchao

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/Nesvilab/philosopher/issues/259#issuecomment-905555134, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AIIMM6ZOUFCKRKBQ2XH3GTLT6T5NXANCNFSM5BSRIL7Q. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email.

---

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues