This is a great tool, I would like to ask how to determine its n_neighbors or n_sampled_neighbors for cosmx data

[CAUiKUN]

No description provided.

[sebastianbirk]

Hello @CAUiKUN,
We recommend to set the "n_neighbors" parameters based on the tissue that you are investigating and, more specifically, the range for which you expect many cell-cell interactions to happen. Empirically, we observed that values between 4 and 8 seem to work very well. The "n_sampled_neighbors" can be left at its default.

[CAUiKUN]

Thank you very much for your reply！

…

---- Replied Message ---- | From | Sebastian ***@***.***> | | Date | 2/24/2025 18:39 | | To | ***@***.***> | | Cc | ***@***.***>, ***@***.***> | | Subject | Re: [Lotfollahi-lab/nichecompass] This is a great tool, I would like to ask how to determine its n_neighbors or n_sampled_neighbors for cosmx data (Issue #85) | Hello @CAUiKUN, We recommend to set the "n_neighbors" parameters based on the tissue that you are investigating and, more specifically, the range for which you expect many cell-cell interactions to happen. Empirically, we observed that values between 4 and 8 seem to work very well. The "n_sampled_neighbors" can be left at its default. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***> sebastianbirk left a comment (Lotfollahi-lab/nichecompass#85) Hello @CAUiKUN, We recommend to set the "n_neighbors" parameters based on the tissue that you are investigating and, more specifically, the range for which you expect many cell-cell interactions to happen. Empirically, we observed that values between 4 and 8 seem to work very well. The "n_sampled_neighbors" can be left at its default. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***>