{
  "id": 546195,
  "title": "understanding of data?",
  "url": "/competitions/czii-cryo-et-object-identification/discussion/546195",
  "author_name": "Manav2805",
  "post_date": "2024-11-14T12:05:47.199000",
  "votes": 6,
  "comment_count": 10,
  "views": 0,
  "content": "<p>is my understanding correct about the data that 0,1 and 2 inside zarr files are just the same files at different resolutions ?</p>",
  "messages": [
    {
      "id": 3045346,
      "postDate": "2024-11-14T12:05:47.200Z",
      "content": "<p>is my understanding correct about the data that 0,1 and 2 inside zarr files are just the same files at different resolutions ?</p>",
      "rawMarkdown": "is my understanding correct about the data that 0,1 and 2 inside zarr files are just the same files at different resolutions ?",
      "votes": 5
    },
    {
      "id": 3045434,
      "postDate": "2024-11-14T13:32:14.030Z",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/manav2805\" target=\"_blank\">@manav2805</a>, that is correct. The Zarr files contain a multiscale image pyramid. The resolutions are defined in the <code>.zattrs</code> files in the <code>*.zarr</code> directories according to v0.4 of the <a href=\"https://ngff.openmicroscopy.org/0.4/index.html\" target=\"_blank\">OME-NGFF specification</a>. Pixel sizes are approximately 10 A, 20 A and 40 A, please refer to the <code>.zattrs</code> for exact scaling factors.</p>",
      "rawMarkdown": "Hi @manav2805, that is correct. The Zarr files contain a multiscale image pyramid. The resolutions are defined in the `.zattrs` files in the `*.zarr` directories according to v0.4 of the [OME-NGFF specification](https://ngff.openmicroscopy.org/0.4/index.html). Pixel sizes are approximately 10 A, 20 A and 40 A, please refer to the `.zattrs` for exact scaling factors.",
      "votes": 3,
      "replies": [
        {
          "id": 3048633,
          "postDate": "2024-11-18T07:49:59.490Z",
          "content": "<p>Thanks <a href=\"https://www.kaggle.com/uermel\" target=\"_blank\">@uermel</a> for your response I have another question we can use anyone of them (0,1,2 inside zarr files) to train the model but lets say if I am using '0' resolution I need to first convert it into 1A from 10A. ? or If I am using '1' i need to convert 1A from 20A ? Is my understanding is correct ?</p>",
          "rawMarkdown": "Thanks @uermel for your response I have another question we can use anyone of them (0,1,2 inside zarr files) to train the model but lets say if I am using '0' resolution I need to first convert it into 1A from 10A. ? or If I am using '1' i need to convert 1A from 20A ? Is my understanding is correct ?",
          "votes": 1,
          "isDeleted": true,
          "replies": [
            {
              "id": 3048663,
              "postDate": "2024-11-18T08:42:21.260Z",
              "content": "<p>If you are using the highest resolution, you multiply by the number of pixels by 10 to get to the Angstrom coordinate system.  At that resolution 1 pixel = 10 Angstroms.  (Technically you multiply by 10.012…and then you would want to add 5.006 to get to the center of the pixel, but that probably doesn't matter for most cases.  The worst case error is less than a pixel in each direction.)</p>",
              "rawMarkdown": "If you are using the highest resolution, you multiply by the number of pixels by 10 to get to the Angstrom coordinate system.  At that resolution 1 pixel = 10 Angstroms.  (Technically you multiply by 10.012...and then you would want to add 5.006 to get to the center of the pixel, but that probably doesn't matter for most cases.  The worst case error is less than a pixel in each direction.)",
              "votes": 2
            },
            {
              "id": 3048691,
              "postDate": "2024-11-18T09:23:04Z",
              "content": "<p>Thanks <a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a> for quick response I got it now. Basically ('0' , '1' ,'2) are High,Med,Low resolution Correct? </p>\n<p>One more question what is the meaning of z coordinate <a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a>  <a href=\"https://www.kaggle.com/uermel\" target=\"_blank\">@uermel</a> ?</p>\n<p>Example : (<strong>184, 630, 630</strong>) from <strong>apo-ferritin</strong> (Took High Resolution '0') , Now I need to convert it into 1A so need to multiply with 10.012 and now it became <strong>(1840,6300,6300)</strong> approximately .Here 1840 tells you number of slide and (6300,6300) are Height and width.In JSON File we are given coordinates like <strong>'location': {'x': 5674.694, 'y': 1114.354, 'z': 565.068}</strong> does it means I need to look into 565 slide and at coordinate** (x:5674.694, y:1114.354)** need to draw sphere of 10A radius (According to host apo-ferritin having 10A radius) ??</p>\n<p>Sorry for naive question hope I manage to elaborate properly.</p>",
              "rawMarkdown": "Thanks @davidlist for quick response I got it now. Basically ('0' , '1' ,'2) are High,Med,Low resolution Correct? \n\nOne more question what is the meaning of z coordinate @davidlist  @uermel ?\n\nExample : (**184, 630, 630**) from **apo-ferritin** (Took High Resolution '0') , Now I need to convert it into 1A so need to multiply with 10.012 and now it became **(1840,6300,6300)** approximately .Here 1840 tells you number of slide and (6300,6300) are Height and width.In JSON File we are given coordinates like **'location': {'x': 5674.694, 'y': 1114.354, 'z': 565.068}** does it means I need to look into 565 slide and at coordinate** (x:5674.694, y:1114.354)** need to draw sphere of 10A radius (According to host apo-ferritin having 10A radius) ??\n\nSorry for naive question hope I manage to elaborate properly.",
              "votes": 1,
              "isDeleted": true
            },
            {
              "id": 3048705,
              "postDate": "2024-11-18T09:38:40.077Z",
              "content": "<p>Yes, to your first question.</p>\n<p>For your second question, think of these as three dimensional images.  Each slide or layer is another pixel in depth.  Slide 0 is closest to you and each subsequent slide is another pixel away.  In your example, it would actually be slide 56 not 565.  The same scaling is used in the z direction as the others.</p>",
              "rawMarkdown": "Yes, to your first question.\n\nFor your second question, think of these as three dimensional images.  Each slide or layer is another pixel in depth.  Slide 0 is closest to you and each subsequent slide is another pixel away.  In your example, it would actually be slide 56 not 565.  The same scaling is used in the z direction as the others."
            },
            {
              "id": 3048723,
              "postDate": "2024-11-18T09:48:51.833Z",
              "content": "<p>Got it Sorry I forget to do the scaling . </p>\n<p><a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a> So <strong>'location': {'x': 5674.694, 'y': 1114.354, 'z': 565.068}</strong> means look into slide 56 and at (567.4694, 'y': 111.4354} we will find apo-ferritin particle having 10A radius . Correct ?</p>\n<p>Thanks for quick response <a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a> </p>",
              "rawMarkdown": "Got it Sorry I forget to do the scaling . \n\n@davidlist So **'location': {'x': 5674.694, 'y': 1114.354, 'z': 565.068}** means look into slide 56 and at (567.4694, 'y': 111.4354} we will find apo-ferritin particle having 10A radius . Correct ?\n\nThanks for quick response @davidlist ",
              "votes": 1,
              "isDeleted": true
            },
            {
              "id": 3048736,
              "postDate": "2024-11-18T10:05:12.650Z",
              "content": "<p>Yes, look at <a href=\"https://www.kaggle.com/code/davidlist/experiment-ts-6-4-visualization\" target=\"_blank\">https://www.kaggle.com/code/davidlist/experiment-ts-6-4-visualization</a> which does exactly that to identify specific particles.</p>\n<p><img src=\"https://www.kaggleusercontent.com/kf/206719764/eyJhbGciOiJkaXIiLCJlbmMiOiJBMTI4Q0JDLUhTMjU2In0..RFaDglrm5gsah04zFvug5Q.7dwCWnBJ8ivyaq53jxruoMObPu415E1dQGiJuCqhrpegc1fY_Baw7SRuknBw-gSAjrN-7aYfcYQPe91bMyOaBglFXworXLxLhrgjFYSFlnqEQ6n77K7327FKEyPDWJYf6BogpJrzovO7lPYVd2b7hz1kzkrP71tn-nObDg9jS6obAjV_0ubeoFFPkww5iK3hI_4gBzktZCoO-2t6zPihx792iJ6TaZCGuYASUdi1_mhqQLILs2-nXRtc0xUOj_Is7sf7CjWQpanbLmz0BtUk40DzylRupI5EWtkReC_ttjsuvm1L7IrAMjTwUYgj_En9Pn9684SrrxAdFug7F6TLZz1nHXcJz0-UyvdsaTvaVqT1BlBtvMJSpKXupn9pIO9ROeYT4TuVYdcu8RNyhKYRJkompXU1DmFh0RAvQC4-2WBjrdRWCrfAFQ2iDlJCDlYEXASNxIl6X9VL1xNJSlhKcIApTZLrQkn7d055hzuFGMKrmxWt45bYT_7H2Qh5NGr1fxv2Y74FRyQDSZSHM6MJ60AFlWQltKsXzQas0ekz9XxXuGVHD_maPOovxw8ThCDGmd48-n5HRR2vm4-nXpgQFmXT1e790ZbsDY_AJiS6B9p50zXlW0O2wE5q0DOP2XW4hcjR6lnhK022-0zM4cGZGV8yNbjY8zhzwu6vUCFowGs.5-jvC3qP7NXhuNP4q2QSrw/__results___files/__results___38_0.png\" alt=\"\"></p>",
              "rawMarkdown": "Yes, look at [https://www.kaggle.com/code/davidlist/experiment-ts-6-4-visualization](https://www.kaggle.com/code/davidlist/experiment-ts-6-4-visualization) which does exactly that to identify specific particles.\n\n![](https://www.kaggleusercontent.com/kf/206719764/eyJhbGciOiJkaXIiLCJlbmMiOiJBMTI4Q0JDLUhTMjU2In0..RFaDglrm5gsah04zFvug5Q.7dwCWnBJ8ivyaq53jxruoMObPu415E1dQGiJuCqhrpegc1fY_Baw7SRuknBw-gSAjrN-7aYfcYQPe91bMyOaBglFXworXLxLhrgjFYSFlnqEQ6n77K7327FKEyPDWJYf6BogpJrzovO7lPYVd2b7hz1kzkrP71tn-nObDg9jS6obAjV_0ubeoFFPkww5iK3hI_4gBzktZCoO-2t6zPihx792iJ6TaZCGuYASUdi1_mhqQLILs2-nXRtc0xUOj_Is7sf7CjWQpanbLmz0BtUk40DzylRupI5EWtkReC_ttjsuvm1L7IrAMjTwUYgj_En9Pn9684SrrxAdFug7F6TLZz1nHXcJz0-UyvdsaTvaVqT1BlBtvMJSpKXupn9pIO9ROeYT4TuVYdcu8RNyhKYRJkompXU1DmFh0RAvQC4-2WBjrdRWCrfAFQ2iDlJCDlYEXASNxIl6X9VL1xNJSlhKcIApTZLrQkn7d055hzuFGMKrmxWt45bYT_7H2Qh5NGr1fxv2Y74FRyQDSZSHM6MJ60AFlWQltKsXzQas0ekz9XxXuGVHD_maPOovxw8ThCDGmd48-n5HRR2vm4-nXpgQFmXT1e790ZbsDY_AJiS6B9p50zXlW0O2wE5q0DOP2XW4hcjR6lnhK022-0zM4cGZGV8yNbjY8zhzwu6vUCFowGs.5-jvC3qP7NXhuNP4q2QSrw/__results___files/__results___38_0.png)",
              "votes": 1
            },
            {
              "id": 3049126,
              "postDate": "2024-11-18T18:26:26.023Z",
              "content": "<p>Thanks for Great Notebook <a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a> </p>\n<p>I understand this code </p>\n<pre><code>\nplt.title()\nplt.imshow(z_ts_6_4[][], cmap=, vmin=-, vmax=)\n_ = plt.scatter([/], [/], edgecolor=, facecolor=)\n</code></pre>\n<p>But For Other view ( Straight-On , Side View and Top View ) I am not getting it .</p>\n<p>For Straight-On </p>\n<pre><code>plt.title()\nplt.imshow(z_ts_6_4[][, :, :], cmap=)\n</code></pre>\n<p>How did you slice 163:207 for Y coordinate and 79:123 for X coordinate .</p>",
              "rawMarkdown": "Thanks for Great Notebook @davidlist \n\nI understand this code \n\n```python\n# {'x': 1019.831, 'y': 1859.831, 'z': 400.424}\nplt.title('Original')\nplt.imshow(z_ts_6_4[0][40], cmap='gray', vmin=-0.00005, vmax=0.00005)\n_ = plt.scatter([1019.831/10], [1859.831/10], edgecolor='red', facecolor='none')\n```\n\nBut For Other view ( Straight-On , Side View and Top View ) I am not getting it .\n\nFor Straight-On \n```python\nplt.title('Straight On')\nplt.imshow(z_ts_6_4[0][40, 163:207, 79:123], cmap='gray')\n```\n\nHow did you slice 163:207 for Y coordinate and 79:123 for X coordinate .",
              "isDeleted": true
            },
            {
              "id": 3049151,
              "postDate": "2024-11-18T19:10:53.503Z",
              "content": "<p>79:123 is columns which corresponds to X.  Dividing X by 10 gives you 101 then add and subtract 22 from 101 to get the range of X to display as an image.  Looking again, that should probably be \"subtract 22 and add 23\" because of how array math works, but it's pretty close.  Then do the same for Y.</p>",
              "rawMarkdown": "79:123 is columns which corresponds to X.  Dividing X by 10 gives you 101 then add and subtract 22 from 101 to get the range of X to display as an image.  Looking again, that should probably be \"subtract 22 and add 23\" because of how array math works, but it's pretty close.  Then do the same for Y."
            }
          ]
        }
      ]
    },
    {
      "id": 3045761,
      "postDate": "2024-11-14T19:06:19.433Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 3045434,
      "author_name": "Utz Ermel",
      "author_url": "",
      "post_date": "2024-11-14T13:32:14.030000",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/manav2805\" target=\"_blank\">@manav2805</a>, that is correct. The Zarr files contain a multiscale image pyramid. The resolutions are defined in the <code>.zattrs</code> files in the <code>*.zarr</code> directories according to v0.4 of the <a href=\"https://ngff.openmicroscopy.org/0.4/index.html\" target=\"_blank\">OME-NGFF specification</a>. Pixel sizes are approximately 10 A, 20 A and 40 A, please refer to the <code>.zattrs</code> for exact scaling factors.</p>",
      "votes": 3,
      "replies": [
        {
          "id": 3048633,
          "author_name": "",
          "author_url": "",
          "post_date": "2024-11-18T07:49:59.490000",
          "content": "<p>Thanks <a href=\"https://www.kaggle.com/uermel\" target=\"_blank\">@uermel</a> for your response I have another question we can use anyone of them (0,1,2 inside zarr files) to train the model but lets say if I am using '0' resolution I need to first convert it into 1A from 10A. ? or If I am using '1' i need to convert 1A from 20A ? Is my understanding is correct ?</p>",
          "votes": 1,
          "replies": [
            {
              "id": 3048663,
              "author_name": "David List",
              "author_url": "",
              "post_date": "2024-11-18T08:42:21.260000",
              "content": "<p>If you are using the highest resolution, you multiply by the number of pixels by 10 to get to the Angstrom coordinate system.  At that resolution 1 pixel = 10 Angstroms.  (Technically you multiply by 10.012…and then you would want to add 5.006 to get to the center of the pixel, but that probably doesn't matter for most cases.  The worst case error is less than a pixel in each direction.)</p>",
              "votes": 2,
              "replies": []
            },
            {
              "id": 3048691,
              "author_name": "",
              "author_url": "",
              "post_date": "2024-11-18T09:23:04",
              "content": "<p>Thanks <a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a> for quick response I got it now. Basically ('0' , '1' ,'2) are High,Med,Low resolution Correct? </p>\n<p>One more question what is the meaning of z coordinate <a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a>  <a href=\"https://www.kaggle.com/uermel\" target=\"_blank\">@uermel</a> ?</p>\n<p>Example : (<strong>184, 630, 630</strong>) from <strong>apo-ferritin</strong> (Took High Resolution '0') , Now I need to convert it into 1A so need to multiply with 10.012 and now it became <strong>(1840,6300,6300)</strong> approximately .Here 1840 tells you number of slide and (6300,6300) are Height and width.In JSON File we are given coordinates like <strong>'location': {'x': 5674.694, 'y': 1114.354, 'z': 565.068}</strong> does it means I need to look into 565 slide and at coordinate** (x:5674.694, y:1114.354)** need to draw sphere of 10A radius (According to host apo-ferritin having 10A radius) ??</p>\n<p>Sorry for naive question hope I manage to elaborate properly.</p>",
              "votes": 1,
              "replies": []
            },
            {
              "id": 3048705,
              "author_name": "David List",
              "author_url": "",
              "post_date": "2024-11-18T09:38:40.077000",
              "content": "<p>Yes, to your first question.</p>\n<p>For your second question, think of these as three dimensional images.  Each slide or layer is another pixel in depth.  Slide 0 is closest to you and each subsequent slide is another pixel away.  In your example, it would actually be slide 56 not 565.  The same scaling is used in the z direction as the others.</p>",
              "votes": 0,
              "replies": []
            },
            {
              "id": 3048723,
              "author_name": "",
              "author_url": "",
              "post_date": "2024-11-18T09:48:51.833000",
              "content": "<p>Got it Sorry I forget to do the scaling . </p>\n<p><a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a> So <strong>'location': {'x': 5674.694, 'y': 1114.354, 'z': 565.068}</strong> means look into slide 56 and at (567.4694, 'y': 111.4354} we will find apo-ferritin particle having 10A radius . Correct ?</p>\n<p>Thanks for quick response <a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a> </p>",
              "votes": 1,
              "replies": []
            },
            {
              "id": 3048736,
              "author_name": "David List",
              "author_url": "",
              "post_date": "2024-11-18T10:05:12.650000",
              "content": "<p>Yes, look at <a href=\"https://www.kaggle.com/code/davidlist/experiment-ts-6-4-visualization\" target=\"_blank\">https://www.kaggle.com/code/davidlist/experiment-ts-6-4-visualization</a> which does exactly that to identify specific particles.</p>\n<p><img src=\"https://www.kaggleusercontent.com/kf/206719764/eyJhbGciOiJkaXIiLCJlbmMiOiJBMTI4Q0JDLUhTMjU2In0..RFaDglrm5gsah04zFvug5Q.7dwCWnBJ8ivyaq53jxruoMObPu415E1dQGiJuCqhrpegc1fY_Baw7SRuknBw-gSAjrN-7aYfcYQPe91bMyOaBglFXworXLxLhrgjFYSFlnqEQ6n77K7327FKEyPDWJYf6BogpJrzovO7lPYVd2b7hz1kzkrP71tn-nObDg9jS6obAjV_0ubeoFFPkww5iK3hI_4gBzktZCoO-2t6zPihx792iJ6TaZCGuYASUdi1_mhqQLILs2-nXRtc0xUOj_Is7sf7CjWQpanbLmz0BtUk40DzylRupI5EWtkReC_ttjsuvm1L7IrAMjTwUYgj_En9Pn9684SrrxAdFug7F6TLZz1nHXcJz0-UyvdsaTvaVqT1BlBtvMJSpKXupn9pIO9ROeYT4TuVYdcu8RNyhKYRJkompXU1DmFh0RAvQC4-2WBjrdRWCrfAFQ2iDlJCDlYEXASNxIl6X9VL1xNJSlhKcIApTZLrQkn7d055hzuFGMKrmxWt45bYT_7H2Qh5NGr1fxv2Y74FRyQDSZSHM6MJ60AFlWQltKsXzQas0ekz9XxXuGVHD_maPOovxw8ThCDGmd48-n5HRR2vm4-nXpgQFmXT1e790ZbsDY_AJiS6B9p50zXlW0O2wE5q0DOP2XW4hcjR6lnhK022-0zM4cGZGV8yNbjY8zhzwu6vUCFowGs.5-jvC3qP7NXhuNP4q2QSrw/__results___files/__results___38_0.png\" alt=\"\"></p>",
              "votes": 1,
              "replies": []
            },
            {
              "id": 3049126,
              "author_name": "",
              "author_url": "",
              "post_date": "2024-11-18T18:26:26.023000",
              "content": "<p>Thanks for Great Notebook <a href=\"https://www.kaggle.com/davidlist\" target=\"_blank\">@davidlist</a> </p>\n<p>I understand this code </p>\n<pre><code>\nplt.title()\nplt.imshow(z_ts_6_4[][], cmap=, vmin=-, vmax=)\n_ = plt.scatter([/], [/], edgecolor=, facecolor=)\n</code></pre>\n<p>But For Other view ( Straight-On , Side View and Top View ) I am not getting it .</p>\n<p>For Straight-On </p>\n<pre><code>plt.title()\nplt.imshow(z_ts_6_4[][, :, :], cmap=)\n</code></pre>\n<p>How did you slice 163:207 for Y coordinate and 79:123 for X coordinate .</p>",
              "votes": 0,
              "replies": []
            },
            {
              "id": 3049151,
              "author_name": "David List",
              "author_url": "",
              "post_date": "2024-11-18T19:10:53.503000",
              "content": "<p>79:123 is columns which corresponds to X.  Dividing X by 10 gives you 101 then add and subtract 22 from 101 to get the range of X to display as an image.  Looking again, that should probably be \"subtract 22 and add 23\" because of how array math works, but it's pretty close.  Then do the same for Y.</p>",
              "votes": 0,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 3045761,
      "author_name": "",
      "author_url": "",
      "post_date": "2024-11-14T19:06:19.433000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3045346": "is my understanding correct about the data that 0,1 and 2 inside zarr files are just the same files at different resolutions ?",
    "3045434": "Hi @manav2805, that is correct. The Zarr files contain a multiscale image pyramid. The resolutions are defined in the `.zattrs` files in the `*.zarr` directories according to v0.4 of the [OME-NGFF specification](https://ngff.openmicroscopy.org/0.4/index.html). Pixel sizes are approximately 10 A, 20 A and 40 A, please refer to the `.zattrs` for exact scaling factors.",
    "3045761": ""
  }
}