Informations on Path file your/ID/file/path.tsv

[mletexier-cnrgh]

Dear All, Thank you so much for sharing.

I have generate the SNVcatalogue file with all my samples:

But for the ID catalogue, when I use the SigProfilerMatrixGenerator script, I have severals output files for each ID:

Each ID file is composed of:

The path file your/ID/file/path.tsv is made of one ? So how I combined all ID files ? Should I do "rbind" all ID files to obtain only one file ?

Thank you in advance for your reply Mélanie

[mletexier-cnrgh]

I think I understand, SigProfilerMatrixGenerator generates severals profils depends on context, so I must take the *SBS83.all file. Sorry for the inconvenience Mélanie

[shadibasyuni]

Sorry for the late reply. I understand your issues have now been resolved? Let me know if you have any outstanding questions

Shadi

Shadi Basyuni / MB BChir, BDS, MRCS, MFDS

Clinical Research Fellow / Nik-Zainal Lab (Early Cancer Institute)Specialty Registrar - Oral & Maxillo-Facial Surgery / Cambridge University HospitalsBye-Fellow / Wolfson College, University of Cambridge @.**@.> / @.**@.>

On 7 Oct 2024, at 16:21, mletexier-cnrgh @.***> wrote:

I think I understand, SigProfilerMatrixGenerator generates severals profils depends on context, so I must take the *SBS96.all file. Sorry for the inconvenience Mélanie

— Reply to this email directly, view it on GitHubhttps://github.com/Nik-Zainal-Group/FFPE_impact/issues/2#issuecomment-2397231693, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AXCJV7Q6S4UN6NA4S2U77BDZ2KRHJAVCNFSM6AAAAABPQEO43GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGOJXGIZTCNRZGM. You are receiving this because you are subscribed to this thread.Message ID: @.***>

[mletexier-cnrgh]

Yes, thank you Shadi.

I would like to ask a more general question. I did obtain the FFPEimpact score for my samples and the results are in line with what I observed on the signatures. However, as a negative control, I also put in FF samples and it turns out that for each FF (5 patients), I have a non-zero FFPE ID, which implies a FFPEimpact score. This score is extremely low, but still higher than some FFPEs. Do you also get a score for your FF samples? For the method, I got the vcf files from Mutect for which I applied the PASS filter on the mutations and nothing more. Do I need to add anything?

Thank you very much for your feedback Mélanie

[shadibasyuni]

Hi Melanie,

It’s great to hear that your FFPEimpact scores align with your observed signatures.

Regarding your FF samples, obtaining a non-zero FFPE ID for them is not entirely unexpected. Running FFPEimpact and FFPE artefact fitting on FF samples can force the mathematical computation for signature fitting, particularly when it comes to indels. The existing COSMIC channels for indels are less granular than SBS, which can (in particular) lead to samples with ID2 and/or ID12 being attributed to ID FFPE (similar profiles). Does this look like what you're observing?

We normally also filter out indels >100bp. Regarding your FFPE samples, I would be interested to know if the overall signature patterns look reasonable? Or do you sense that there may be underlying artefacts not accounted for by FFPEimpact? In our paper, we acknowledge that more signatures may emerge with increased sample analysis, particularly as new artefacts could be introduced at different stages of DNA extraction or library preparation. We're currently exploring this ourselves with a new cohort, where we’ve noticed a potential novel artefactual signature linked to their prep methods.

Please don’t hesitate to reach out if you'd like to discuss this further or have any other questions!

Best regards,

Shadi

Shadi Basyuni / MB BChir, BDS, MRCS, MFDS

Clinical Research Fellow / Nik-Zainal Lab (Early Cancer Institute) Specialty Registrar - Oral & Maxillo-Facial Surgery / Cambridge University Hospitals Bye-Fellow / Wolfson College, University of Cambridge

@. @.> / @. @.>

On 14 Oct 2024, at 11:40, mletexier-cnrgh @.***> wrote:

Yes, thank you Shadi.

I would like to ask a more general question. I did obtain the FFPEimpact score for my samples and the results are in line with what I observed on the signatures. However, as a negative control, I also put in FF samples and it turns out that for each FF (5 patients), I have a non-zero FFPE ID, which implies a FFPEimpact score. This score is extremely low, but still higher than some FFPEs. Do you also get a score for your FF samples? For the method, I got the vcf files from Mutect for which I applied the PASS filter on the mutations and nothing more. Do I need to add anything?

Thank you very much for your feedback Mélanie

— Reply to this email directly, view it on GitHub https://github.com/Nik-Zainal-Group/FFPE_impact/issues/2#issuecomment-2410803118, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXCJV7XTHT3E554M26YQHBLZ3ONRDAVCNFSM6AAAAABPQEO43GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMJQHAYDGMJRHA. You are receiving this because you commented.

[mletexier-cnrgh]

Dear Shadi, My apologies for the delay in replying.

For FF samples, after excluding indels greater than 100bp, I still obtain a FFPEimpact score. Below are the results of the IDs identified upon removing the ID_FFPE to initiate the Fit analysis. Notably, there is no detection of ID2 or ID12. Upon reintroducing the ID_FFPE, I lose ID10; I have not examined its profile, but it is possible that this signature closely resembles the ID_FFPE. Additionally, for two patients, ID5 also becomes absent.

For FFPE samples, despite the SBS_FFPE signature being notably weak in our samples, we are able to identify the signatures of our FF. We tested two extraction kits, and our observations indicate that the SBS37 signature is present exclusively in FFPE samples extracted with Qiagen. Have you noticed a similar phenomenon?

Sincerely Mélanie