Same peptide has different razor proteins from Philosopher

[jd690764]

Describe the bug
A clear and concise description of what the bug is.
Hi,
I am uploading the log file.
log_2020-08-08_09-19-30.txt

Any help would be greatly appreciated.

Thanks,
Janos

---

If you're submitting a bug report, please attach log file

The log file can be saved from FragPipe:

• by clicking the Export Log button on the Run tab.
• or just copy text from the output console on Run tab to a text file.

[fcyu]

Hi @jd690764

The warnings from IonQuant looks weird. Can you share all psm.tsv files to us? I can send a link for you to upload.

Thanks,

Fengchao

[jd690764]

Hi Fengchao, Thank you for your help. Yes, I can do that. Please, send me the link to upload. Janos

…

On Mon, Aug 10, 2020 at 12:42 PM Fengchao ***@***.***> wrote: Hi @jd690764 <https://github.com/jd690764> The warnings from IonQuant looks weird. Can you share all psm.tsv files to us? I can send a link for you to upload. Thanks, Fengchao — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#218 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACP7QY2WTWFL5ZT5IEYBG2TSABET7ANCNFSM4P2IVE5Q> .

[fcyu]

Hi Janos,

Thanks. Please upload your files to https://umich.app.box.com/f/bc3183ac9ec44c318c2cd43fa3f36e32

Best,

Fengchao

[jd690764]

Hi Fengchao, I uploaded them in a zip file. Thank you, Janos

…

On Tue, Aug 11, 2020 at 1:33 PM Fengchao ***@***.***> wrote: Hi Janos, Thanks. Please upload your files to https://umich.app.box.com/f/bc3183ac9ec44c318c2cd43fa3f36e32 Best, Fengchao — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#218 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACP7QY2MU7PDX6VRIWSNU6TSAGTJJANCNFSM4P2IVE5Q> .

[fcyu]

Thanks Janos,

I checked one peptide HGYIGEFEIIDDHR in psm_1.tsv and psm_2.tsv, and it has two different razor proteins:

In psm_1.tsv:

From psm_2.tsv:

If you process this data in one batch, there should be a bug in Philosopher.

@prvst Can you take a look when you have time? The psm files are in https://umich.box.com/s/1iw3iumyf1uq489ygz3dh9l3aoa33m3i

Thanks,

Fengchao

[prvst]

How were these tables generated? Are these from the same experiment? Different data sets? Is it a combined analysis?

[fcyu]

There is a log here (#218 (comment)). Look like all runs were processed together.

[prvst]

I suggest using the latest version of Philosopher

[anesvi]

Also, can you try with philosopher generated UniProt database. Download it with FragPioe. Philosopher sometimes has issues with custom databases ( although RefSeq should be fine)

…

Sent from my iPhone On Aug 12, 2020, at 9:07 AM, Felipe Leprevost <notifications@github.com> wrote:  External Email - Use Caution I suggest using the latest version of Philosopher — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#218 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AIIMM62Y6SRW5GTC2ATFUO3SAKHTPANCNFSM4P2IVE5Q>. ********************************************************** Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

[jd690764]

Yes, these were run as a single batch. I actually ran this twice and it gave the exact same error. I have used the same library in the past and had no issues with it. I'll repeat the run with the newest version of philosopher and report back to you probably tomorrow or Friday. Thank you, Janos

…

On Wed, Aug 12, 2020 at 6:05 AM Felipe Leprevost ***@***.***> wrote: I suggest using the latest version of Philosopher — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#218 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACP7QY4HVMVAUOU2UWOO2KLSAKHTNANCNFSM4P2IVE5Q> .

[jd690764]

Hi, I updated philosopher to the latest version (3.2.9) and recreated the protein library. I used a smaller dataset that doesn't take as much time to run, but it produced the same error. Here is one peptide that produced these conflicts: 1: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009922 NP_001009562

2: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009562 NP_001009608 3: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009562 NP_001009608 4: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009922 NP_001009608 5: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009562 NP_001009925 6: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009562 NP_001009925 7: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009562 NP_001009182 8: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009562 NP_001009182 9: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009562 NP_001009922 10: TFNQVEIKPEMIGHYLGEFSITYKPVK NP_001009562 NP_001009922 > t[V1 == 'TFNQVEIKPEMIGHYLGEFSITYKPVK', 2:3] %>% unlist %>% unique %>% sort [1] "NP_001009182" "NP_001009562" "NP_001009608" "NP_001009922" "NP_001009925"

All these ids are valid, they are in the library. But the peptide is only present in 2 proteins: grep -n 'TFNQVEIKPEMIGHYLGEFSITYKPVK'

2020-08-14-decoys-contam-covid2_human_proteome.fa 3936:MAEVEQKKKRTFRKFTYRGVDLDQLLDMSYEQLMQLYSARQRRRLNRGLRRKQHSLLKRLRKAKKEAPPMEKPEVVKTHLRDMIILPEMVGSMVGVYNGKTFNQVEIKPEMIGHYLGEFSITYKPVKHGRPGIGATHSSRFIPLK 40048:MLGRGADLAEVEQKKKRTFRKFTYRGVDLDQLLDMSYEQLMQLYSARQRRRLNRGLRRKQHSLLKRLRKAKKEAPPMEKPEVVKTHLRDMIILPEMVGSMVGVYNGKTFNQVEIKPEMIGHYLGEFSITYKPVKHGRPGIGATHSSRFIPLK

more +3935 2020-08-14-decoys-contam-covid2_human_proteome.fa|head -1

>NP_001009 | 40S ribosomal protein S15 isoform 2 | 6209 | RPS15 | 145 | P62841 | sp

more +40047 2020-08-14-decoys-contam-covid2_human_proteome.fa|head -1

>NP_001295155 | 40S ribosomal protein S15 isoform 1 | 6209 | RPS15 | 152 | K7ELC2 | tr

It seems the real protein id should be NP_001009, but for some reason these 5 other ids - of the form NP_001009nnn - are assigned (there are a total of 65 ids of this form in the library, plus the correct NP_001009). I looked at a few other sequences giving warnings, they all have this same issue of a shorter id replaced by ids with 3 additional digits appended to the correct one. I am not sure what this means, but hopefully it makes some sense to you. Thank you for looking into this. Janos

…

On Wed, Aug 12, 2020 at 12:02 PM Janos Demeter ***@***.***> wrote: Yes, these were run as a single batch. I actually ran this twice and it gave the exact same error. I have used the same library in the past and had no issues with it. I'll repeat the run with the newest version of philosopher and report back to you probably tomorrow or Friday. Thank you, Janos On Wed, Aug 12, 2020 at 6:05 AM Felipe Leprevost ***@***.***> wrote: > I suggest using the latest version of Philosopher > > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub > <#218 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/ACP7QY4HVMVAUOU2UWOO2KLSAKHTNANCNFSM4P2IVE5Q> > . >

[prvst]

@jd690764

Could you share with me all your reports and the database? I also need you to confirm the commands you executed in the last run.

[jd690764]

Hi Felipe, here are the files: https://office365stanford-my.sharepoint.com/:f:/g/personal/jdemeter_stanford_edu/Ej3jarHuaatHlpvqr9f2OGEBC_zKorCAR5Z8XFRMECwj2A?e=LHwa2b Let me know if you need anything else. Thank you, Janos

…

On Mon, Aug 17, 2020 at 5:57 AM Felipe Leprevost ***@***.***> wrote: @jd690764 <https://github.com/jd690764> Could you share with me all your reports and the database? I also need you to confirm the commands you executed in the last run. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#218 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACP7QY6OMNEQLIAXNSCHEVLSBESNHANCNFSM4P2IVE5Q> .

[prvst]

How did you get this database?

[jd690764]

I have my own library that I made manually and then ran the latest philosopher according to the instructions online. Janos

…

On Mon, Aug 17, 2020 at 11:25 AM Felipe Leprevost ***@***.***> wrote: How did you get this database? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#218 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACP7QYZE3JVGGQGKTF7CAE3SBFYY3ANCNFSM4P2IVE5Q> .

[prvst]

I suspect that there might be an issue with the parsing of the FASTA sequences. I suggest that you take a look at our documentation on how to work with the pre-existing database, and how to format it for Philosopher properly.

[fcyu]

@prvst I think one possible reason may be Philosopher matching substring rather than a whole string of protein IDs. That is why it messes up NP_001009 with all other proteins starting with NP_001009 (e.g., NP_001009922, NP_001009562, NP_001009608).

[prvst]

Yes, it could be. But we need to be sure first that the formatting is OK, that way we can eliminate it from the possible problems.

[jd690764]

Hi Felipe, I looked at the library again - before adding the contaminants - and it seems correct to me. (The library I sent you is a derivative of a library I successfully used before, but which is also giving the same error now - see my original report.) After adding the contaminants it also looks good, the only differences are the added contaminants and reformatting of the protein sequences (removing the line breaks that were present after every 60 amino acids). Also, it is recognized as correct by fragpipe. Thanks, Janos

…

On Mon, Aug 17, 2020 at 11:42 AM Felipe Leprevost ***@***.***> wrote: Yes, it could be. But we need to be sure first that the formatting is OK, that way we can eliminate it from the possible problems. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#218 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACP7QY4H7TBNU4MNY7VMUXTSBF22HANCNFSM4P2IVE5Q> .

[jd690764]

Hi Felipe, Indeed, the error was due to the library. Description lines like this:

NP_060876 | mucin-4 isoform a precursor | 4585 | MUC4 | 5412 | Q99102 | sp

produce errors, but lines like this:

NP_060876|4585|MUC4 mucin-4 isoform a precursor

don't. The first form of description used to work with earlier versions of msfragger/philosopher. Thank you for your help, Janos

…

On Mon, Aug 17, 2020 at 12:22 PM jd690764 ***@***.***> wrote: Hi Felipe, I looked at the library again - before adding the contaminants - and it seems correct to me. (The library I sent you is a derivative of a library I successfully used before, but which is also giving the same error now - see my original report.) After adding the contaminants it also looks good, the only differences are the added contaminants and reformatting of the protein sequences (removing the line breaks that were present after every 60 amino acids). Also, it is recognized as correct by fragpipe. Thanks, Janos On Mon, Aug 17, 2020 at 11:42 AM Felipe Leprevost < ***@***.***> wrote: > Yes, it could be. But we need to be sure first that the formatting is OK, > that way we can eliminate it from the possible problems. > > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub > <#218 (comment) >, > or unsubscribe > < https://github.com/notifications/unsubscribe-auth/ACP7QY4H7TBNU4MNY7VMUXTSBF22HANCNFSM4P2IVE5Q > > . > — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#218 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACP7QY73TWFMU7FEYXHMH43SBF7PBANCNFSM4P2IVE5Q> .

[prvst]

I'm glad that you were able to find the issue.