Inconsistent number of channels when using DeepCell/Mesmer segmentation

[yiqiaochen7]

Hello, I was trying to run the pipeline with example data and when I entered "steinbock segment deepcell --app mesmer --minmax". It showed ValueError: Input data must have 2 channels. Input data only has 47 channels. It seems to work when I opened the "panel.csv" file in "steinbock new" folder and change the "keep" value from 1 to 0 for the first 46 rows (left the keep value as 1 for Pt196 and Pb206). Is it feasible? Would it affect the following results?

Best wishes, Yiqiao

[jwindhager]

Hi @yiqiaochen7

Some clarifications:

• The keep column specifies which channels are retained during image export, see here and here. If there is no keep column present in the panel file, all channels will be retained. Do not alter the keep column after image preprocessing.
• The deepcell column determines nuclear (1) and cytoplasmic (2) channels, see here and here. DeepCell/Mesmer segmentation requires at least one channel marked as nuclear and at least one channel marked as cytoplasmic.

So, to solve your issue, please do not alter the keep column, but adapt the deepcell column instead. In case you are following the workshop material, a suitable panel file was provided with the example data, see here.

Hope this helps.

[yiqiaochen7]

Hi Jonas,

Thank you very much for the reply. It worked. May I ask one more question? Is there link introducing how to process my raw data (.mcd file) so that I can run it with the steinbock pipeline. Should I make it to zip? should I generate that panel.csv file?

Best wishes, Yiqiao

---

From: Jonas Windhager @.> Sent: 24 January 2023 21:30 To: BodenmillerGroup/steinbock @.> Cc: Yiqiao Chen @.>; Mention @.> Subject: Re: [BodenmillerGroup/steinbock] ValueError when running "5. Segmenting cells using DeeCell/Mesmer" (Issue #163)

Hi @yiqiaochen7https://github.com/yiqiaochen7

Some clarifications:

• The keep column specifies which channels are retained during image export, see herehttps://bodenmillergroup.github.io/steinbock/latest/file-types/ and herehttps://bodenmillergroup.github.io/steinbock/latest/cli/preprocessing/#imaging-mass-cytometry-imc. If there is no keep column present in the panel file, all channels will be retained. Do not alter the keep column after image preprocessing.
• The deepcell column determines nuclear (1) and cytoplasmic (2) channels, see herehttps://bodenmillergroup.github.io/steinbock/latest/file-types/ and herehttps://bodenmillergroup.github.io/steinbock/latest/cli/segmentation/#deepcell. DeepCell/Mesmer segmentation requires at least one channel marked as nuclear and at least one channel marked as cytoplasmic.

So, to solve your issue, please do not alter the keep column, but adapt the deepcell column instead. In case you are following the workshop material, a suitable panel file was provided with the example data, see herehttps://github.com/BodenmillerGroup/ImagingWorkshop2023/tree/main/data.

Hope this helps.

— Reply to this email directly, view it on GitHubhttps://github.com/BodenmillerGroup/steinbock/issues/163#issuecomment-1401953805, or unsubscribehttps://github.com/notifications/unsubscribe-auth/A5LCUNKJJUPSTB7D2QND63TWT7KRHANCNFSM6AAAAAAUFCSSZA. You are receiving this because you were mentioned.Message ID: @.***>

[jwindhager]

Please consult the steinbock documentation