Thanks for the organizers and Kaggle for that remarkable competition !
\nIt gives us access to cutting-edge technology datasets which are related to many interesting unresolved biological research questions. Understanding such questions might (in some future) lead to new insights for drug design, disease treatments and so on.
So, some my colleagues and I are willing to organize some activity around (quite widely around) that competition.
\nImportant: it is not about achieving top score on the LB, but rather about research and education. With a possible goal - writing a joint review/benchmarking/original results paper on the basis of that activity.
\nJoin us - notify under that message or contact me directly or join our Telegram group: https://t.me/sberlogacompete
What is the plan:
\n1) Organize free regular webinars on biological and data science aspects related to the dataset to share experience and networking. So we are looking for experts who can share their experience.
\n2) List some biologically interesting research questions which can be explored with that and similar datasets. Organize kind of \"pet-projects\" around that tasks.
\nSee discussion for details: https://www.kaggle.com/competitions/open-problems-multimodal/discussion/348295
3) List the research papers relevant to the subject. Organize \"journal club\" discussions around them. Try to analyze and \"benchmark\" the proposed methods using that and related datasets.
\n4) If activity is successful write we can write a joint review/benchmarking paper on the basis of that activity.
\nSimilar to what Andrei Zinovyev and me have done recently: https://arxiv.org/abs/2208.05229 on single cell RNA-seq cell cycle analysis. Note - almost all the work has been done on Kaggle - you can find almost all datasets and code on Kaggle: https://www.kaggle.com/alexandervc/datasets
What we have:
\n1) The first webinars are already arranged everybody is welcome to join - the zoom link will be posted here shortly before meetings start:
\n👨🔬 Andrew Lukyanenko (Kaggle Grandmaster) \"Getting started with Kaggle\"
\n⌚️ Monday 29 August , 17.00 (Paris Time)
\nGoogle calendar link
https://us02web.zoom.us/j/85275922838?pwd=MTAxZHVoYnEycmNXNW0vUDNWMkp2UT09
\n👨🔬 Larisa Okorokova (Université Côte d'Azur) \"Introduction to the biological story behind the Open problems in single cell analysis 2022 competition\"
\n⌚️ Wednesday 31 August , 18.00 (Paris Time)
\nGoogle calendar link\n
(The video records will be stored on our youtube channel https://www.youtube.com/c/SciBerloga ):
\n2) In our paper mentioned above we have listed several open questions related to cell cycle analysis for the single cell data. In case someone is interested we can arrange some activity in that direction, but hopefully there should be other topics from other people.
\n3) We have made preliminary discussion in our community \"Sberloga\" - there are many biological background people, who are willing to improve their data science skills - and that Kaggle competition is the best place to try,
\nand there are many data science people who are willing to improve their biology knowledge - that Kaggle competition is the best place to try.
\nYou are welcome to join the Telegram chat https://t.me/sberlogacompete which might be useful for short discussions.
Of course, we will follow all the competition rules, in particular \"No private sharing\".
\nRoughly speaking activity might hopefully be similar to standard activity of a research unit - small groups leaded by seniors, work on the their topics proposed by seniors,
\non the other hand having common \"kitchen\" where everybody is discussing and exchanging experience, and hopefully finally producing some joint paper - but everything is going on fully virtually and without formal obligations.
\nLet us look how it will work. We are looking for every one who is willing to participate in the activity at any level of involvement.
Thanks for the organizers and Kaggle for that remarkable competition !
\nIt gives us access to cutting-edge technology datasets which are related to many interesting unresolved biological research questions. Understanding such questions might (in some future) lead to new insights for drug design, disease treatments and so on.
So, some my colleagues and I are willing to organize some activity around (quite widely around) that competition.
\nImportant: it is not about achieving top score on the LB, but rather about research and education. With a possible goal - writing a joint review/benchmarking/original results paper on the basis of that activity.
\nJoin us - notify under that message or contact me directly or join our Telegram group: https://t.me/sberlogacompete
What is the plan:
\n1) Organize free regular webinars on biological and data science aspects related to the dataset to share experience and networking. So we are looking for experts who can share their experience.
\n2) List some biologically interesting research questions which can be explored with that and similar datasets. Organize kind of \"pet-projects\" around that tasks.
\nSee discussion for details: https://www.kaggle.com/competitions/open-problems-multimodal/discussion/348295
3) List the research papers relevant to the subject. Organize \"journal club\" discussions around them. Try to analyze and \"benchmark\" the proposed methods using that and related datasets.
\n4) If activity is successful write we can write a joint review/benchmarking paper on the basis of that activity.
\nSimilar to what Andrei Zinovyev and me have done recently: https://arxiv.org/abs/2208.05229 on single cell RNA-seq cell cycle analysis. Note - almost all the work has been done on Kaggle - you can find almost all datasets and code on Kaggle: https://www.kaggle.com/alexandervc/datasets
What we have:
\n1) The first webinars are already arranged everybody is welcome to join - the zoom link will be posted here shortly before meetings start:
\n👨🔬 Andrew Lukyanenko (Kaggle Grandmaster) \"Getting started with Kaggle\"
\n⌚️ Monday 29 August , 17.00 (Paris Time)
\nGoogle calendar link
https://us02web.zoom.us/j/85275922838?pwd=MTAxZHVoYnEycmNXNW0vUDNWMkp2UT09
\n👨🔬 Larisa Okorokova (Université Côte d'Azur) \"Introduction to the biological story behind the Open problems in single cell analysis 2022 competition\"
\n⌚️ Wednesday 31 August , 18.00 (Paris Time)
\nGoogle calendar link\n
(The video records will be stored on our youtube channel https://www.youtube.com/c/SciBerloga ):
\n2) In our paper mentioned above we have listed several open questions related to cell cycle analysis for the single cell data. In case someone is interested we can arrange some activity in that direction, but hopefully there should be other topics from other people.
\n3) We have made preliminary discussion in our community \"Sberloga\" - there are many biological background people, who are willing to improve their data science skills - and that Kaggle competition is the best place to try,
\nand there are many data science people who are willing to improve their biology knowledge - that Kaggle competition is the best place to try.
\nYou are welcome to join the Telegram chat https://t.me/sberlogacompete which might be useful for short discussions.
Of course, we will follow all the competition rules, in particular \"No private sharing\".
\nRoughly speaking activity might hopefully be similar to standard activity of a research unit - small groups leaded by seniors, work on the their topics proposed by seniors,
\non the other hand having common \"kitchen\" where everybody is discussing and exchanging experience, and hopefully finally producing some joint paper - but everything is going on fully virtually and without formal obligations.
\nLet us look how it will work. We are looking for every one who is willing to participate in the activity at any level of involvement.
Hello, thank you very much for this initiative Alexander. I plan on starting to learn graph neural networks and I was wondering if they would be applicable to this problem and this dataset.
", "rawMarkdown": "Hello, thank you very much for this initiative Alexander. I plan on starting to learn graph neural networks and I was wondering if they would be applicable to this problem and this dataset.", "votes": null }, { "id": "1921412", "postDate": "08/31/2022 20:21:12", "content": "@marcyane7 Hello Dayane , thank you very much for you kind words !
\nYes, there are some applications of GNN to similar tasks - but not sure exactly the same:
\nI do not remember the research papers off-hand , but if you google \"single cell graph neural network\" - there should be (I think) several dozens.
There is an interesting paper from famous Stanford Jure Leskovec:
\nhttps://www.biorxiv.org/content/10.1101/2022.07.12.499735v1
\nhttps://twitter.com/yusufroohani/status/1547965695744360448
\nI started to look on it, but than that comptetion appeared - and I changed on it.
\nThere are GNNs , the task is NOT exactly the same, but still in some sense related.
\nI started to download datasets to Kaggle which were used in that paper
\nhttps://www.kaggle.com/datasets/alexandervc/scrnaseq-crisprperturbseqnormanselectedpart
\nI plan to return to that topic later - it is very very interesting !
2 years ago as now
", "rawMarkdown": "2 years ago as now", "votes": null } ], "comments": [ { "id": 1917266, "author_name": "marcyane7", "author_url": "", "post_date": "08/28/2022 15:33:46", "content": "Hello, thank you very much for this initiative Alexander. I plan on starting to learn graph neural networks and I was wondering if they would be applicable to this problem and this dataset.
", "votes": null, "replies": [ { "id": 1921412, "author_name": "alexandervc", "author_url": "", "post_date": "08/31/2022 20:21:12", "content": "@marcyane7 Hello Dayane , thank you very much for you kind words !
\nYes, there are some applications of GNN to similar tasks - but not sure exactly the same:
\nI do not remember the research papers off-hand , but if you google \"single cell graph neural network\" - there should be (I think) several dozens.
There is an interesting paper from famous Stanford Jure Leskovec:
\nhttps://www.biorxiv.org/content/10.1101/2022.07.12.499735v1
\nhttps://twitter.com/yusufroohani/status/1547965695744360448
\nI started to look on it, but than that comptetion appeared - and I changed on it.
\nThere are GNNs , the task is NOT exactly the same, but still in some sense related.
\nI started to download datasets to Kaggle which were used in that paper
\nhttps://www.kaggle.com/datasets/alexandervc/scrnaseq-crisprperturbseqnormanselectedpart
\nI plan to return to that topic later - it is very very interesting !
2 years ago as now
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1916323": "Thanks for the organizers and Kaggle for that remarkable competition !\nIt gives us access to cutting-edge technology datasets which are related to many interesting unresolved biological research questions. Understanding such questions might (in some future) lead to new insights for drug design, disease treatments and so on.\n\nSo, some my colleagues and I are willing to organize some activity around (quite widely around) that competition. \nImportant: it is not about achieving top score on the LB, but rather about research and education. With a possible goal - writing a joint review/benchmarking/original results paper on the basis of that activity. \n**Join us** - notify under that message or contact me directly or join our Telegram group: https://t.me/sberlogacompete\n\n**What is the plan:** \n\n1) Organize free regular webinars on biological and data science aspects related to the dataset to share experience and networking. So we are looking for experts who can share their experience. \n\n2) List some biologically interesting research questions which can be explored with that and similar datasets. Organize kind of \"pet-projects\" around that tasks. \nSee discussion for details: https://www.kaggle.com/competitions/open-problems-multimodal/discussion/348295\n\n3) List the research papers relevant to the subject. Organize \"journal club\" discussions around them. Try to analyze and \"benchmark\" the proposed methods using that and related datasets. \n\n4) If activity is successful write we can write a joint review/benchmarking paper on the basis of that activity. \nSimilar to what Andrei Zinovyev and me have done recently: https://arxiv.org/abs/2208.05229 on single cell RNA-seq cell cycle analysis. Note - almost all the work has been done on Kaggle - you can find almost all datasets and code on Kaggle: https://www.kaggle.com/alexandervc/datasets\n\n**What we have:** \n\n1) The first webinars are already arranged everybody is welcome to join - the zoom link will be posted here shortly before meetings start:\n\n👨🔬 Andrew Lukyanenko (Kaggle Grandmaster) \"Getting started with Kaggle\"\n⌚️ Monday 29 August , 17.00 (Paris Time)\n[Google calendar link](https://calendar.google.com/calendar/render?action=TEMPLATE&dates=20220829T150000Z%2F20220829T173000Z&details=Zoom%20link%20will%20be%20available%20in%20%0Ahttps%3A%2F%2Ft.me%2Fsberlogacompete%0ADetailed%20announcement%20will%20be%20given%20in%20%0Ahttps%3A%2F%2Ft.me%2Fsberlogabig%0A&text=%40SBERLOGABIG%20Andrew%20Lukyanenko%20%28Kaggle%20Grandmaster%29%20%22Getting%20started%20with%20Kaggle%22)\n\n\nhttps://us02web.zoom.us/j/85275922838?pwd=MTAxZHVoYnEycmNXNW0vUDNWMkp2UT09\n\n\n\n👨🔬 Larisa Okorokova (Université Côte d'Azur) \"Introduction to the biological story behind the Open problems in single cell analysis 2022 competition\"\n⌚️ Wednesday 31 August , 18.00 (Paris Time)\n[Google calendar link\n](https://calendar.google.com/calendar/render?action=TEMPLATE&dates=20220831T160000Z%2F20220831T173000Z&details=Zoom%20link%20will%20be%20available%20in%20%0Ahttps%3A%2F%2Ft.me%2Fsberlogacompete%0ADetailed%20announcement%20will%20be%20given%20in%20%0Ahttps%3A%2F%2Ft.me%2Fsberlogabig%0A&text=%40SBERLOGABIG%20Larisa%20Okorokova%20%28Universit%C3%A9%20C%C3%B4te%20d%27Azur%29%20%22Introduction%20to%20the%20biological%20story%20behind%20%20the%20%20Open%20problems%20in%20single%20cell%20analysis%202022%20competition%22)\n\n(The video records will be stored on our youtube channel https://www.youtube.com/c/SciBerloga ): \n\n2) In our paper mentioned above we have listed several open questions related to cell cycle analysis for the single cell data. In case someone is interested we can arrange some activity in that direction, but hopefully there should be other topics from other people.\n\n3) We have made preliminary discussion in our community \"Sberloga\" - there are many biological background people, who are willing to improve their data science skills - and that Kaggle competition is the best place to try,\nand there are many data science people who are willing to improve their biology knowledge - that Kaggle competition is the best place to try.\nYou are welcome to join the Telegram chat https://t.me/sberlogacompete which might be useful for short discussions. \n\nOf course, we will follow all the competition rules, in particular \"No private sharing\". \n\nRoughly speaking activity might hopefully be similar to standard activity of a research unit - small groups leaded by seniors, work on the their topics proposed by seniors,\non the other hand having common \"kitchen\" where everybody is discussing and exchanging experience, and hopefully finally producing some joint paper - but everything is going on fully virtually and without formal obligations. \nLet us look how it will work. We are looking for every one who is willing to participate in the activity at any level of involvement.", "1917266": "Hello, thank you very much for this initiative Alexander. I plan on starting to learn graph neural networks and I was wondering if they would be applicable to this problem and this dataset.", "1921412": "marcyane7 Hello Dayane , thank you very much for you kind words !\nYes, there are some applications of GNN to similar tasks - but not sure exactly the same:\nI do not remember the research papers off-hand , but if you google \"single cell graph neural network\" - there should be (I think) several dozens. \n\nThere is an interesting paper from famous Stanford Jure Leskovec:\nhttps://www.biorxiv.org/content/10.1101/2022.07.12.499735v1\nhttps://twitter.com/yusufroohani/status/1547965695744360448\nI started to look on it, but than that comptetion appeared - and I changed on it. \nThere are GNNs , the task is NOT exactly the same, but still in some sense related. \nI started to download datasets to Kaggle which were used in that paper \nhttps://www.kaggle.com/datasets/alexandervc/scrnaseq-crisprperturbseqnormanselectedpart\nI plan to return to that topic later - it is very very interesting !", "2978388": "2 years ago as now" }, "source": "meta" }