Closed elliotmgray closed 6 years ago
Hi Eliot,
Thanks for the report. I believe that the clustering is actually happening with Euclidean distance but the 'Angular Distance' is mistakenly displayed in the clustering result list. One way to tell - Do you get a different result if you were to try X-shift with Angular distance using the same K on the same dataset?
I will fix it in the next release
Nikolay On Fri, Jun 23, 2017 at 1:46 PM Elliot Gray notifications@github.com wrote:
Hello,
I am using Vortex to analyze cell data which is always positive, and for which "zero" has the specific meaning of "no signal". Therefore, I wish to use the Euclidean Distance measure, rather than Angular Distance.
The problem is this: when I select "Euclidean Distance", the algorithm runs, and then the message that sows up next to each data set in the "clustering" window under the "Distance Measure" column says "Angular Distance". In other words, it looks like the "Euclidean Distance" button does not work and instead defaults to "Angular Distance".
Best, Elliot
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/nolanlab/vortex/issues/6, or mute the thread https://github.com/notifications/unsubscribe-auth/ADacL_7orVC-EFGWcAaiMFacmJ-BavRZks5sHCQPgaJpZM4OEBvY .
-- Nikolay
This message is intended for the named addressee(s) only. It may contain
privileged and confidential information and protected by a copyright. Any disclosure, copying or distribution of this message is prohibited and may be unlawful. If you are not the intended recipient, please destroy this message and notify me immediately. Thank you for your cooperation.
Dear Nikolay,
Thank you for your response.
I can confirm that running the algorithm with the different distance metric selections does give different results. I suppose you are probably right that the correct methods are being used, but simply failing to display the correct text in the summary.
You can see the attached images if you like: force directed layouts for the algorithm applied to the same dataset, k=20, with both metrics. (I also checked the cluster ID's for verification of the identity of the two clusterings.
Best, Elliot
On Mon, Jun 26, 2017 at 8:29 PM, nsamusik notifications@github.com wrote:
Hi Eliot,
Thanks for the report. I believe that the clustering is actually happening with Euclidean distance but the 'Angular Distance' is mistakenly displayed in the clustering result list. One way to tell - Do you get a different result if you were to try X-shift with Angular distance using the same K on the same dataset?
I will fix it in the next release
Nikolay On Fri, Jun 23, 2017 at 1:46 PM Elliot Gray notifications@github.com wrote:
Hello,
I am using Vortex to analyze cell data which is always positive, and for which "zero" has the specific meaning of "no signal". Therefore, I wish to use the Euclidean Distance measure, rather than Angular Distance.
The problem is this: when I select "Euclidean Distance", the algorithm runs, and then the message that sows up next to each data set in the "clustering" window under the "Distance Measure" column says "Angular Distance". In other words, it looks like the "Euclidean Distance" button does not work and instead defaults to "Angular Distance".
Best, Elliot
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/nolanlab/vortex/issues/6, or mute the thread https://github.com/notifications/unsubscribe-auth/ADacL_7orVC- EFGWcAaiMFacmJ-BavRZks5sHCQPgaJpZM4OEBvY .
-- Nikolay
This message is intended for the named addressee(s) only. It may contain
privileged and confidential information and protected by a copyright. Any disclosure, copying or distribution of this message is prohibited and may be unlawful. If you are not the intended recipient, please destroy this message and notify me immediately. Thank you for your cooperation.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/nolanlab/vortex/issues/6#issuecomment-311242907, or mute the thread https://github.com/notifications/unsubscribe-auth/AQ27kd9dB3yR2otv41XqZscg2nAYwrzzks5sIHcrgaJpZM4OEBvY .
Dear Elliot,
Please check out the new release - Vortex 29-Jun-2019
The issue should be fixed
Best,
Nikolay
Hello,
I am using Vortex to analyze cell data which is always positive, and for which "zero" has the specific meaning of "no signal". Therefore, I wish to use the Euclidean Distance measure, rather than Angular Distance.
The problem is this: when I select "Euclidean Distance", the algorithm runs, and then the message that sows up next to each data set in the "clustering" window under the "Distance Measure" column says "Angular Distance". In other words, it looks like the "Euclidean Distance" button does not work and instead defaults to "Angular Distance".
Best, Elliot