{
  "id": 462534,
  "title": "118th Place Solution for the Open Problems – Single-Cell Perturbations Competition draft",
  "url": "/competitions/open-problems-single-cell-perturbations/discussion/462534",
  "author_name": "",
  "post_date": "2023-12-20T11:19:05.251000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I would like to express gratitude to Kaggle for hosting the competition and Open Problems in Single-Cell Analysis scientific collaboration for a single-cell perturbational dataset and for select 144 compounds from the Library of Integrated Network-Based Cellular Signatures (LINCS) Connectivity Map dataset (<a href=\"https://pubmed.ncbi.nlm.nih.gov/29195078/\" target=\"_blank\">PMID: 29195078</a>) <a href=\"https://www.cell.com/action/showPdf?pii=S0092-8674%2817%2931309-0\" target=\"_blank\"> L1000</a></p>\n<h1>1.    Integration of biological knowledge</h1>\n<h2>Context</h2>\n<p>• Business context: <a href=\"https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/overview\" target=\"_blank\">https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/overview</a><br>\n• Data context: <a href=\"https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/data\" target=\"_blank\">https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/data</a></p>\n<h1>2.    Exploration of the problem</h1>\n<p><a href=\"https://ru.wikipedia.org/wiki/%D0%9A%D0%BE%D0%BB%D0%B8%D1%87%D0%B5%D1%81%D1%82%D0%B2%D0%B5%D0%BD%D0%BD%D1%8B%D0%B9_%D0%B0%D0%BD%D0%B0%D0%BB%D0%B8%D0%B7_%D1%8D%D0%BA%D1%81%D0%BF%D1%80%D0%B5%D1%81%D1%81%D0%B8%D0%B8_%D0%B3%D0%B5%D0%BD%D0%BE%D0%B2\" target=\"_blank\">Gene expression profiling</a></p>\n<p>The wiki contains models such as the Generalized Linear Model, which is used by the … place :</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F3e1cc01e2e176b1984ddebf6537d029b%2FIMG_20231220_155835.jpg?generation=1703080145948652&amp;alt=media\"></p>\n<p>Train</p>\n<table>\n<thead>\n<tr>\n<th>Compound</th>\n<th>Gene 0</th>\n<th>1</th>\n<th>..</th>\n<th>18 211</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>1, ABT-199 (GDC-0199)</td>\n<td>2547</td>\n<td>2006</td>\n<td>..</td>\n<td>3387</td>\n</tr>\n<tr>\n<td>2, ABT737</td>\n<td>780</td>\n<td>381</td>\n<td>..</td>\n<td>2093</td>\n</tr>\n<tr>\n<td>3, AMD-070 (hydrochloride)</td>\n<td>3335</td>\n<td>3424</td>\n<td>..</td>\n<td>1355</td>\n</tr>\n<tr>\n<td>..</td>\n<td>..</td>\n<td>..</td>\n<td>..</td>\n<td>..</td>\n</tr>\n<tr>\n<td>613,Myeloid cells,YK 4-279</td>\n<td>3373</td>\n<td>1433</td>\n<td>..</td>\n<td>1618</td>\n</tr>\n</tbody>\n</table>\n<p>Inference</p>\n<table>\n<thead>\n<tr>\n<th>Compound</th>\n<th>Gene 0</th>\n<th>1</th>\n<th>..</th>\n<th>18 211</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>254,Myeloid cells,YK 4-279</td>\n<td>?</td>\n<td>?</td>\n<td>?</td>\n<td>?</td>\n</tr>\n</tbody>\n</table>\n<p>Predict: signed -log10(p-values)</p>\n<h1>3.    Model design</h1>\n<p>3.1 Gene order model<br>\n3.2 Plot line model<br>\n3.3 Marker point on plot model<br>\n3.4 Multiplier model</p>\n<h3>3.1 Gene order model</h3>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F66047a0cc4015b2b6e3280bf543d6a68%2F2023-12-20%20%2015.34.41.png?generation=1703075796711416&amp;alt=media\"></p>\n<p>Search Queries: 19,B cell, BMS-387032<br>\nSearch Results: </p>\n<table>\n<thead>\n<tr>\n<th>Gene</th>\n<th>Rank 0 .. 18 211</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>AL1173282</td>\n<td>2547</td>\n</tr>\n<tr>\n<td>AC2397982</td>\n<td>2006</td>\n</tr>\n<tr>\n<td>AC0118992</td>\n<td>682</td>\n</tr>\n<tr>\n<td>AP0056711</td>\n<td>3470</td>\n</tr>\n<tr>\n<td>…</td>\n<td>…</td>\n</tr>\n<tr>\n<td>ACAP1</td>\n<td>2072</td>\n</tr>\n<tr>\n<td>ARHGAP15</td>\n<td>3590</td>\n</tr>\n</tbody>\n</table>\n<p>Predict rank. Not predict value.</p>\n<h3>3.2 Plot line model</h3>\n<p>Curve class prediction Attachments compound.pdf :</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F59437c342447b7260b22e2020552a7b0%2F2023-12-20%20%2018.47.27.png?generation=1703088550144370&amp;alt=media\"></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F9f5120b3fbb36e68e40ac8feef6dfa0e%2F2023-12-20%20%2018.45.03.png?generation=1703088579086403&amp;alt=media\"></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2Fc043c2cd488217adf1a1fabb95f71dbe%2F2023-12-20%20%2018.40.33.png?generation=1703088612243004&amp;alt=media\"></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F805ea55569935020bf6bb5a1e68b03b6%2F2023-12-20%20%2018.53.40.png?generation=1703088628570177&amp;alt=media\"></p>\n<h3>3.3 Multiplier model</h3>\n<p>Predict point:</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F859c65d626ecc5312a34953294675b9a%2F2023-12-20%20%2019.11.37.png?generation=1703089042788063&amp;alt=media\"></p>\n<p>Predict R for points.</p>\n<h3>3.4 Multiplier model</h3>\n<p>Predict zoom:<br>\nMagic of 4th place:</p>\n<p>Examine the multiplier coefficient in the attachments: Multiplier.xlsx</p>\n<table>\n<thead>\n<tr>\n<th>Gene</th>\n<th>Cell</th>\n<th>Multiplier</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>BMS-387032</td>\n<td>B cells</td>\n<td>3 ,09</td>\n</tr>\n<tr>\n<td>Lamivudine</td>\n<td>B cells</td>\n<td>2 ,9</td>\n</tr>\n<tr>\n<td>AZD-8330</td>\n<td>Myeloid cells</td>\n<td>2 ,031</td>\n</tr>\n<tr>\n<td>Perhexiline</td>\n<td>Myeloid cells</td>\n<td>1 ,870</td>\n</tr>\n<tr>\n<td>AT13387</td>\n<td>Myeloid cells</td>\n<td>1 ,675</td>\n</tr>\n</tbody>\n</table>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2Ffb4c98e53f29150d2bbe8c6cb587e618%2F2023-12-20%20%2021.47.14.png?generation=1703098200209837&amp;alt=media\"></p>\n<h2>Overview of the approach</h2>\n<h2>Data preprocessing, feature engineering:</h2>\n<h2>The models</h2>\n<h2>Validation Strategy</h2>\n<h2>Details of the submission</h2>\n<h2>What was impactful about the submission.</h2>\n<h2>What was tried and didn’t work.</h2>\n<h1>4.    Robustness</h1>\n<table>\n<thead>\n<tr>\n<th>Model</th>\n<th>Private</th>\n<th>Public</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td></td>\n<td></td>\n<td></td>\n</tr>\n</tbody>\n</table>\n<h1>5.    Documentation &amp; code style</h1>\n<h2>Code samples feature engineering:</h2>\n<h2>Code samples model training:</h2>\n<h2>Code samples model inference:</h2>\n<h1>6.    Reproducibility</h1>\n<table>\n<thead>\n<tr>\n<th>name</th>\n<th>link</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>118th Place notebook</td>\n<td><a href=\"https://www.kaggle.com/emmawilsonev/118th-place-solution-for-the-open-problems-singl\" target=\"_blank\">https://www.kaggle.com/emmawilsonev/118th-place-solution-for-the-open-problems-singl</a></td>\n</tr>\n</tbody>\n</table>\n<h2>Helpful notebooks:</h2>",
  "messages": [
    {
      "id": 2568263,
      "postDate": "2023-12-20T11:19:05.250Z",
      "content": "<p>I would like to express gratitude to Kaggle for hosting the competition and Open Problems in Single-Cell Analysis scientific collaboration for a single-cell perturbational dataset and for select 144 compounds from the Library of Integrated Network-Based Cellular Signatures (LINCS) Connectivity Map dataset (<a href=\"https://pubmed.ncbi.nlm.nih.gov/29195078/\" target=\"_blank\">PMID: 29195078</a>) <a href=\"https://www.cell.com/action/showPdf?pii=S0092-8674%2817%2931309-0\" target=\"_blank\"> L1000</a></p>\n<h1>1.    Integration of biological knowledge</h1>\n<h2>Context</h2>\n<p>• Business context: <a href=\"https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/overview\" target=\"_blank\">https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/overview</a><br>\n• Data context: <a href=\"https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/data\" target=\"_blank\">https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/data</a></p>\n<h1>2.    Exploration of the problem</h1>\n<p><a href=\"https://ru.wikipedia.org/wiki/%D0%9A%D0%BE%D0%BB%D0%B8%D1%87%D0%B5%D1%81%D1%82%D0%B2%D0%B5%D0%BD%D0%BD%D1%8B%D0%B9_%D0%B0%D0%BD%D0%B0%D0%BB%D0%B8%D0%B7_%D1%8D%D0%BA%D1%81%D0%BF%D1%80%D0%B5%D1%81%D1%81%D0%B8%D0%B8_%D0%B3%D0%B5%D0%BD%D0%BE%D0%B2\" target=\"_blank\">Gene expression profiling</a></p>\n<p>The wiki contains models such as the Generalized Linear Model, which is used by the … place :</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F3e1cc01e2e176b1984ddebf6537d029b%2FIMG_20231220_155835.jpg?generation=1703080145948652&amp;alt=media\"></p>\n<p>Train</p>\n<table>\n<thead>\n<tr>\n<th>Compound</th>\n<th>Gene 0</th>\n<th>1</th>\n<th>..</th>\n<th>18 211</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>1, ABT-199 (GDC-0199)</td>\n<td>2547</td>\n<td>2006</td>\n<td>..</td>\n<td>3387</td>\n</tr>\n<tr>\n<td>2, ABT737</td>\n<td>780</td>\n<td>381</td>\n<td>..</td>\n<td>2093</td>\n</tr>\n<tr>\n<td>3, AMD-070 (hydrochloride)</td>\n<td>3335</td>\n<td>3424</td>\n<td>..</td>\n<td>1355</td>\n</tr>\n<tr>\n<td>..</td>\n<td>..</td>\n<td>..</td>\n<td>..</td>\n<td>..</td>\n</tr>\n<tr>\n<td>613,Myeloid cells,YK 4-279</td>\n<td>3373</td>\n<td>1433</td>\n<td>..</td>\n<td>1618</td>\n</tr>\n</tbody>\n</table>\n<p>Inference</p>\n<table>\n<thead>\n<tr>\n<th>Compound</th>\n<th>Gene 0</th>\n<th>1</th>\n<th>..</th>\n<th>18 211</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>254,Myeloid cells,YK 4-279</td>\n<td>?</td>\n<td>?</td>\n<td>?</td>\n<td>?</td>\n</tr>\n</tbody>\n</table>\n<p>Predict: signed -log10(p-values)</p>\n<h1>3.    Model design</h1>\n<p>3.1 Gene order model<br>\n3.2 Plot line model<br>\n3.3 Marker point on plot model<br>\n3.4 Multiplier model</p>\n<h3>3.1 Gene order model</h3>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F66047a0cc4015b2b6e3280bf543d6a68%2F2023-12-20%20%2015.34.41.png?generation=1703075796711416&amp;alt=media\"></p>\n<p>Search Queries: 19,B cell, BMS-387032<br>\nSearch Results: </p>\n<table>\n<thead>\n<tr>\n<th>Gene</th>\n<th>Rank 0 .. 18 211</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>AL1173282</td>\n<td>2547</td>\n</tr>\n<tr>\n<td>AC2397982</td>\n<td>2006</td>\n</tr>\n<tr>\n<td>AC0118992</td>\n<td>682</td>\n</tr>\n<tr>\n<td>AP0056711</td>\n<td>3470</td>\n</tr>\n<tr>\n<td>…</td>\n<td>…</td>\n</tr>\n<tr>\n<td>ACAP1</td>\n<td>2072</td>\n</tr>\n<tr>\n<td>ARHGAP15</td>\n<td>3590</td>\n</tr>\n</tbody>\n</table>\n<p>Predict rank. Not predict value.</p>\n<h3>3.2 Plot line model</h3>\n<p>Curve class prediction Attachments compound.pdf :</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F59437c342447b7260b22e2020552a7b0%2F2023-12-20%20%2018.47.27.png?generation=1703088550144370&amp;alt=media\"></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F9f5120b3fbb36e68e40ac8feef6dfa0e%2F2023-12-20%20%2018.45.03.png?generation=1703088579086403&amp;alt=media\"></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2Fc043c2cd488217adf1a1fabb95f71dbe%2F2023-12-20%20%2018.40.33.png?generation=1703088612243004&amp;alt=media\"></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F805ea55569935020bf6bb5a1e68b03b6%2F2023-12-20%20%2018.53.40.png?generation=1703088628570177&amp;alt=media\"></p>\n<h3>3.3 Multiplier model</h3>\n<p>Predict point:</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F859c65d626ecc5312a34953294675b9a%2F2023-12-20%20%2019.11.37.png?generation=1703089042788063&amp;alt=media\"></p>\n<p>Predict R for points.</p>\n<h3>3.4 Multiplier model</h3>\n<p>Predict zoom:<br>\nMagic of 4th place:</p>\n<p>Examine the multiplier coefficient in the attachments: Multiplier.xlsx</p>\n<table>\n<thead>\n<tr>\n<th>Gene</th>\n<th>Cell</th>\n<th>Multiplier</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>BMS-387032</td>\n<td>B cells</td>\n<td>3 ,09</td>\n</tr>\n<tr>\n<td>Lamivudine</td>\n<td>B cells</td>\n<td>2 ,9</td>\n</tr>\n<tr>\n<td>AZD-8330</td>\n<td>Myeloid cells</td>\n<td>2 ,031</td>\n</tr>\n<tr>\n<td>Perhexiline</td>\n<td>Myeloid cells</td>\n<td>1 ,870</td>\n</tr>\n<tr>\n<td>AT13387</td>\n<td>Myeloid cells</td>\n<td>1 ,675</td>\n</tr>\n</tbody>\n</table>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2Ffb4c98e53f29150d2bbe8c6cb587e618%2F2023-12-20%20%2021.47.14.png?generation=1703098200209837&amp;alt=media\"></p>\n<h2>Overview of the approach</h2>\n<h2>Data preprocessing, feature engineering:</h2>\n<h2>The models</h2>\n<h2>Validation Strategy</h2>\n<h2>Details of the submission</h2>\n<h2>What was impactful about the submission.</h2>\n<h2>What was tried and didn’t work.</h2>\n<h1>4.    Robustness</h1>\n<table>\n<thead>\n<tr>\n<th>Model</th>\n<th>Private</th>\n<th>Public</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td></td>\n<td></td>\n<td></td>\n</tr>\n</tbody>\n</table>\n<h1>5.    Documentation &amp; code style</h1>\n<h2>Code samples feature engineering:</h2>\n<h2>Code samples model training:</h2>\n<h2>Code samples model inference:</h2>\n<h1>6.    Reproducibility</h1>\n<table>\n<thead>\n<tr>\n<th>name</th>\n<th>link</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>118th Place notebook</td>\n<td><a href=\"https://www.kaggle.com/emmawilsonev/118th-place-solution-for-the-open-problems-singl\" target=\"_blank\">https://www.kaggle.com/emmawilsonev/118th-place-solution-for-the-open-problems-singl</a></td>\n</tr>\n</tbody>\n</table>\n<h2>Helpful notebooks:</h2>",
      "rawMarkdown": "I would like to express gratitude to Kaggle for hosting the competition and Open Problems in Single-Cell Analysis scientific collaboration for a single-cell perturbational dataset and for select 144 compounds from the Library of Integrated Network-Based Cellular Signatures (LINCS) Connectivity Map dataset ([PMID: 29195078](https://pubmed.ncbi.nlm.nih.gov/29195078/)) [ L1000](https://www.cell.com/action/showPdf?pii=S0092-8674%2817%2931309-0)\n\n# 1.\tIntegration of biological knowledge\n\n## Context\n• Business context: https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/overview\n• Data context: https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/data\n\n# 2.\tExploration of the problem\n\n[Gene expression profiling](https://ru.wikipedia.org/wiki/%D0%9A%D0%BE%D0%BB%D0%B8%D1%87%D0%B5%D1%81%D1%82%D0%B2%D0%B5%D0%BD%D0%BD%D1%8B%D0%B9_%D0%B0%D0%BD%D0%B0%D0%BB%D0%B8%D0%B7_%D1%8D%D0%BA%D1%81%D0%BF%D1%80%D0%B5%D1%81%D1%81%D0%B8%D0%B8_%D0%B3%D0%B5%D0%BD%D0%BE%D0%B2)\n\nThe wiki contains models such as the Generalized Linear Model, which is used by the ... place :\n \n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F3e1cc01e2e176b1984ddebf6537d029b%2FIMG_20231220_155835.jpg?generation=1703080145948652&alt=media\" width=\"200px\" height=\"230px\">\n\nTrain\n\t\t\t\t\n|Compound\t|Gene 0|\t1|\t..|\t18 211|\n| --- | --- |--- |\n|1, ABT-199 (GDC-0199)|\t2547|\t2006|\t..\t|3387|\n|2, ABT737|\t780\t|381|\t..\t|2093|\n|3, AMD-070 (hydrochloride)|\t3335|\t3424|\t..\t|1355|\n|..|\t..|\t..\t|..\t|..|\n|613,Myeloid cells,YK 4-279|\t3373|\t1433|\t..\t|1618|\n\t\t\t\t\t\t\t\nInference\n\t\t\t\t\n|Compound\t|Gene 0|\t1|\t..|\t18 211|\n| --- | --- |--- |\n|254,Myeloid cells,YK 4-279|\t?|\t?\t|?\t|?|\n\nPredict: signed -log10(p-values)\n\n# 3.\tModel design\n\n3.1 Gene order model\n3.2 Plot line model\n3.3 Marker point on plot model\n3.4 Multiplier model\n\n\n### 3.1 Gene order model\n \n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F66047a0cc4015b2b6e3280bf543d6a68%2F2023-12-20%20%2015.34.41.png?generation=1703075796711416&alt=media\" width=\"400px\" height=\"400px\">\n\nSearch Queries: 19,B cell, BMS-387032\nSearch Results: \n|Gene|                       Rank 0 .. 18 211 |\n| --- | --- |\n|AL1173282|\t\t2547|\n|AC2397982|\t\t2006|\n|AC0118992|\t\t682|\n|AP0056711|\t\t3470|\n| ...| ...|\n|ACAP1|\t\t2072|\n|ARHGAP15|\t\t3590|\n\nPredict rank. Not predict value.\n\n### 3.2 Plot line model\nCurve class prediction Attachments compound.pdf :\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F59437c342447b7260b22e2020552a7b0%2F2023-12-20%20%2018.47.27.png?generation=1703088550144370&alt=media\" width=\"400px\" height=\"360px\">\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F9f5120b3fbb36e68e40ac8feef6dfa0e%2F2023-12-20%20%2018.45.03.png?generation=1703088579086403&alt=media\" width=\"400px\" height=\"360px\">\n\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2Fc043c2cd488217adf1a1fabb95f71dbe%2F2023-12-20%20%2018.40.33.png?generation=1703088612243004&alt=media\" width=\"400px\" height=\"360px\">\n\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F805ea55569935020bf6bb5a1e68b03b6%2F2023-12-20%20%2018.53.40.png?generation=1703088628570177&alt=media\" width=\"400px\" height=\"360px\">\n\n### 3.3 Multiplier model\nPredict point:\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F859c65d626ecc5312a34953294675b9a%2F2023-12-20%20%2019.11.37.png?generation=1703089042788063&alt=media\" width=\"400px\" height=\"480px\">\n\nPredict R for points.\n\n### 3.4 Multiplier model\nPredict zoom:\nMagic of 4th place:\n\nExamine the multiplier coefficient in the attachments: Multiplier.xlsx\n| Gene  | Cell | Multiplier |\n| --- | --- | --- |\n|BMS-387032|\tB cells|\t 3 ,09|\n|Lamivudine|\tB cells|\t 2 ,9|\n|AZD-8330|\tMyeloid cells\t| 2 ,031|\n|Perhexiline|\tMyeloid cells\t| 1 ,870|\n|AT13387|\tMyeloid cells\t| 1 ,675|\n\n \n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2Ffb4c98e53f29150d2bbe8c6cb587e618%2F2023-12-20%20%2021.47.14.png?generation=1703098200209837&alt=media\" width=\"400px\" height=\"520px\">\n\n## Overview of the approach\n\n## Data preprocessing, feature engineering:\n\n## The models\n\n## Validation Strategy\n\n## Details of the submission\n\n## What was impactful about the submission.\n\n## What was tried and didn’t work.\n\n# 4.\tRobustness\n\n\n| Model | Private | Public |\n| --- | --- |--- |\n|  |  |  |\n\n\n# 5.\tDocumentation & code style\n\n## Code samples feature engineering:\n\n## Code samples model training:\n\n## Code samples model inference:\n\n\n\n# 6.\tReproducibility\n\n|  name |  link |\n| --- | --- |\n|  118th Place notebook  | https://www.kaggle.com/emmawilsonev/118th-place-solution-for-the-open-problems-singl | \n\n## Helpful notebooks:\n",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2568263": "I would like to express gratitude to Kaggle for hosting the competition and Open Problems in Single-Cell Analysis scientific collaboration for a single-cell perturbational dataset and for select 144 compounds from the Library of Integrated Network-Based Cellular Signatures (LINCS) Connectivity Map dataset ([PMID: 29195078](https://pubmed.ncbi.nlm.nih.gov/29195078/)) [ L1000](https://www.cell.com/action/showPdf?pii=S0092-8674%2817%2931309-0)\n\n# 1.\tIntegration of biological knowledge\n\n## Context\n• Business context: https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/overview\n• Data context: https://www.kaggle.com/competitions/open-problems-single-cell-perturbations/data\n\n# 2.\tExploration of the problem\n\n[Gene expression profiling](https://ru.wikipedia.org/wiki/%D0%9A%D0%BE%D0%BB%D0%B8%D1%87%D0%B5%D1%81%D1%82%D0%B2%D0%B5%D0%BD%D0%BD%D1%8B%D0%B9_%D0%B0%D0%BD%D0%B0%D0%BB%D0%B8%D0%B7_%D1%8D%D0%BA%D1%81%D0%BF%D1%80%D0%B5%D1%81%D1%81%D0%B8%D0%B8_%D0%B3%D0%B5%D0%BD%D0%BE%D0%B2)\n\nThe wiki contains models such as the Generalized Linear Model, which is used by the ... place :\n \n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F3e1cc01e2e176b1984ddebf6537d029b%2FIMG_20231220_155835.jpg?generation=1703080145948652&alt=media\" width=\"200px\" height=\"230px\">\n\nTrain\n\t\t\t\t\n|Compound\t|Gene 0|\t1|\t..|\t18 211|\n| --- | --- |--- |\n|1, ABT-199 (GDC-0199)|\t2547|\t2006|\t..\t|3387|\n|2, ABT737|\t780\t|381|\t..\t|2093|\n|3, AMD-070 (hydrochloride)|\t3335|\t3424|\t..\t|1355|\n|..|\t..|\t..\t|..\t|..|\n|613,Myeloid cells,YK 4-279|\t3373|\t1433|\t..\t|1618|\n\t\t\t\t\t\t\t\nInference\n\t\t\t\t\n|Compound\t|Gene 0|\t1|\t..|\t18 211|\n| --- | --- |--- |\n|254,Myeloid cells,YK 4-279|\t?|\t?\t|?\t|?|\n\nPredict: signed -log10(p-values)\n\n# 3.\tModel design\n\n3.1 Gene order model\n3.2 Plot line model\n3.3 Marker point on plot model\n3.4 Multiplier model\n\n\n### 3.1 Gene order model\n \n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F66047a0cc4015b2b6e3280bf543d6a68%2F2023-12-20%20%2015.34.41.png?generation=1703075796711416&alt=media\" width=\"400px\" height=\"400px\">\n\nSearch Queries: 19,B cell, BMS-387032\nSearch Results: \n|Gene|                       Rank 0 .. 18 211 |\n| --- | --- |\n|AL1173282|\t\t2547|\n|AC2397982|\t\t2006|\n|AC0118992|\t\t682|\n|AP0056711|\t\t3470|\n| ...| ...|\n|ACAP1|\t\t2072|\n|ARHGAP15|\t\t3590|\n\nPredict rank. Not predict value.\n\n### 3.2 Plot line model\nCurve class prediction Attachments compound.pdf :\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F59437c342447b7260b22e2020552a7b0%2F2023-12-20%20%2018.47.27.png?generation=1703088550144370&alt=media\" width=\"400px\" height=\"360px\">\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F9f5120b3fbb36e68e40ac8feef6dfa0e%2F2023-12-20%20%2018.45.03.png?generation=1703088579086403&alt=media\" width=\"400px\" height=\"360px\">\n\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2Fc043c2cd488217adf1a1fabb95f71dbe%2F2023-12-20%20%2018.40.33.png?generation=1703088612243004&alt=media\" width=\"400px\" height=\"360px\">\n\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F805ea55569935020bf6bb5a1e68b03b6%2F2023-12-20%20%2018.53.40.png?generation=1703088628570177&alt=media\" width=\"400px\" height=\"360px\">\n\n### 3.3 Multiplier model\nPredict point:\n\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2F859c65d626ecc5312a34953294675b9a%2F2023-12-20%20%2019.11.37.png?generation=1703089042788063&alt=media\" width=\"400px\" height=\"480px\">\n\nPredict R for points.\n\n### 3.4 Multiplier model\nPredict zoom:\nMagic of 4th place:\n\nExamine the multiplier coefficient in the attachments: Multiplier.xlsx\n| Gene  | Cell | Multiplier |\n| --- | --- | --- |\n|BMS-387032|\tB cells|\t 3 ,09|\n|Lamivudine|\tB cells|\t 2 ,9|\n|AZD-8330|\tMyeloid cells\t| 2 ,031|\n|Perhexiline|\tMyeloid cells\t| 1 ,870|\n|AT13387|\tMyeloid cells\t| 1 ,675|\n\n \n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16791068%2Ffb4c98e53f29150d2bbe8c6cb587e618%2F2023-12-20%20%2021.47.14.png?generation=1703098200209837&alt=media\" width=\"400px\" height=\"520px\">\n\n## Overview of the approach\n\n## Data preprocessing, feature engineering:\n\n## The models\n\n## Validation Strategy\n\n## Details of the submission\n\n## What was impactful about the submission.\n\n## What was tried and didn’t work.\n\n# 4.\tRobustness\n\n\n| Model | Private | Public |\n| --- | --- |--- |\n|  |  |  |\n\n\n# 5.\tDocumentation & code style\n\n## Code samples feature engineering:\n\n## Code samples model training:\n\n## Code samples model inference:\n\n\n\n# 6.\tReproducibility\n\n|  name |  link |\n| --- | --- |\n|  118th Place notebook  | https://www.kaggle.com/emmawilsonev/118th-place-solution-for-the-open-problems-singl | \n\n## Helpful notebooks:\n"
  }
}