Add correct zoom areas for gender embedding

[ritagonmar]

After the main gender figure, it zooms-in in some regions of the embedding corresponding to examples of areas with one-gender prevalence. Do you think we could show the examples that we have in the figure from the paper? I could give you the coordinates, because they were not included in the csv you have. I also have several other cool examples (that are not included right now in any figure), in case we want to add some other areas too.

[bmschmidt]

So what those are now are some things I auto-generated. They're in this section of the paper: https://github.com/bmschmidt/pubmed-explorer/blob/main/src/scrollership/pubmed.md?plain=1#L638-L673

Rather than send them to me, do you think you could edit some of that code directly? Each button is a markdown block like below--just change the bbox and the label in that section of the document. I can then clean up any problems that creates in the compiling.

` ``button
label: breastfeeding
api: {"zoom": {"bbox": {"x": [82.0, 84.0], "y": [130.0, 132.0]}}}

` ``

[ritagonmar]

Sure, I will do it!

[dkobak]

FYI I am done with editing the markdown file for now, so if you want to edit, go ahead.

[bmschmidt]

I've just republished with current changes; will take a closer look later.

On Tue, Mar 28, 2023 at 9:20 AM Dmitry Kobak @.***> wrote:

FYI I am done with editing the markdown file for now, so if you want to edit, go ahead.

— Reply to this email directly, view it on GitHub https://github.com/bmschmidt/pubmed-explorer/issues/27#issuecomment-1486876265, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIPFZRJLSBMIKMYFOAALHDW6LQRFANCNFSM6AAAAAAWJBLORA . You are receiving this because you commented.Message ID: @.***>

[ritagonmar]

@bmschmidt , are the coordinates in the interactive version inverted in the y-axis in comparison to the t-SNE embedding coordinates (meaning a point (x_0, y_0) for me would be (x_0, -y_0) for you)? I am asking because the zoom regions I added seem to be wrong.

[ritagonmar]

I changed them accordingly because looking at where it zoomed in and where it should have been, it seemed to me that that was the case. If it is not, the coordinates may need some readjustment.

[bmschmidt]

Yes, I've flipped the sign on all y values because for webGL (and canvas) the origin is the upper left corner.

[dkobak]

I did some edits to the gender part now (and also elsewhere). I am done for today, if you have a moment to republish before our morning, it would be great!