is it possible to use deepminds alphafold2 here ?
", "messages": [ { "id": 2309535, "postDate": "2023-06-19T17:39:40.407Z", "content": "is it possible to use deepminds alphafold2 here ?
", "rawMarkdown": "is it possible to use deepminds alphafold2 here ?", "votes": 3 }, { "id": 2309696, "postDate": "2023-06-19T20:18:22.030Z", "content": "Structure can inform of function, but note that AF2 predictions are also just predictions. Some proteins would have experimental X-ray or NMR structures which you may find at PDB.
\nYou can retrieve AF2 predictions for most of the superset here: https://alphafold.ebi.ac.uk/ (I'm note sure if anyone created this dataset already for this challenge?)
\nMore on prediction of function from structure https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1000485
", "rawMarkdown": "Structure can inform of function, but note that AF2 predictions are also just predictions. Some proteins would have experimental X-ray or NMR structures which you may find at [PDB](https://www.rcsb.org/).\n\nYou can retrieve AF2 predictions for most of the superset here: https://alphafold.ebi.ac.uk/ (I'm note sure if anyone created this dataset already for this challenge?)\n\nMore on prediction of function from structure https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1000485\n\n\n\n\n", "votes": 1, "replies": [ { "id": 2310052, "postDate": "2023-06-20T05:00:15.347Z", "content": "assume i get the structure of the gene , how will i be able to predict its functions? I am going through the links currently .
", "rawMarkdown": "assume i get the structure of the gene , how will i be able to predict its functions? I am going through the links currently .\n", "replies": [ { "id": 2310096, "postDate": "2023-06-20T05:31:33.430Z", "content": "You can do structure alignment search query against training protein, and use KNN to generate the confidence score
", "rawMarkdown": "You can do structure alignment search query against training protein, and use KNN to generate the confidence score", "votes": 1 }, { "id": 2310339, "postDate": "2023-06-20T09:40:17.643Z", "content": "task 1 : Get the structure of the gene
\ntask 2 : structure alignment search query against training protein, and use KNN to generate the confidence score.
\ncan we have discussion about all the steps to be executed. would you like to join @victorfernandezalbor
Chat gpt answers!!!!!
\nOnce trained, AlphaFold can take a protein's sequence of amino acids and predict its shape accurately , and how do i predict its function ?
\nChatGPT
Predicting a protein's function based solely on its sequence and shape is a more complex task compared to predicting its shape alone. While protein shape provides valuable insights, understanding protein function involves considering additional factors such as evolutionary relationships, biochemical experiments, and structural analysis of protein interactions.
\nHere's a simplified explanation of how protein function prediction works:
\nSequence Analysis: The first analyze the protein\n\nHomology Modeling: a protein\n\nStructural Analysis: you mentioned, AlphaFold can predict the shape a protein accurately. Once the protein\n\nFunctional Annotations: Scientists also rely databases existing knowledge annotate assign functions proteins. These databases contain information about experimentally validated functions various proteins. comparing a newly discovered protein proteins known functions, scientists can make predictions about its potential .\n\nExperimental Validation: confirm the predicted , experimental studies are conducted. These experiments might involve laboratory techniques such enzymatic assays, protein-protein interaction assays, genetic manipulations observe the protein\nIt's important to note that predicting protein function is an ongoing area of research, and it can be challenging to accurately predict function solely based on sequence and structure. Many factors, including experimental validation, are crucial for confirming and refining these predictions.
\nOverall, predicting protein function involves a combination of sequence analysis, structural analysis, comparative genomics, and experimental validation. These approaches help scientists make educated predictions about the likely functions of newly discovered proteins.
", "rawMarkdown": "Chat gpt answers!!!!!\nOnce trained, AlphaFold can take a protein's sequence of amino acids and predict its shape accurately , and how do i predict its function ?\nChatGPT\n\nPredicting a protein's function based solely on its sequence and shape is a more complex task compared to predicting its shape alone. While protein shape provides valuable insights, understanding protein function involves considering additional factors such as evolutionary relationships, biochemical experiments, and structural analysis of protein interactions.\n\nHere's a simplified explanation of how protein function prediction works:\n\n Sequence Analysis: The first step is to analyze the protein's amino acid sequence. Scientists compare the sequence to a database of known protein sequences to identify similar proteins that already have known functions. This approach assumes that proteins with similar sequences likely have similar functions.\n\n Homology Modeling: If a protein's sequence closely matches that of a known protein with a known function, scientists can infer that the new protein may have a similar function. This inference is based on the principle that similar sequences often have similar functions.\n\n Structural Analysis: As you mentioned, AlphaFold can predict the shape of a protein accurately. Once the protein's structure is known, scientists can analyze its 3D arrangement to gain insights into its function. Certain structural features, such as active sites or binding pockets, can provide clues about the protein's role in specific biological processes.\n\n Functional Annotations: Scientists also rely on databases and existing knowledge to annotate or assign functions to proteins. These databases contain information about experimentally validated functions for various proteins. By comparing a newly discovered protein to proteins with known functions, scientists can make predictions about its potential function.\n\n Experimental Validation: To confirm the predicted function, experimental studies are conducted. These experiments might involve laboratory techniques such as enzymatic assays, protein-protein interaction assays, or genetic manipulations to observe the protein's impact on cellular processes.\n\nIt's important to note that predicting protein function is an ongoing area of research, and it can be challenging to accurately predict function solely based on sequence and structure. Many factors, including experimental validation, are crucial for confirming and refining these predictions.\n\nOverall, predicting protein function involves a combination of sequence analysis, structural analysis, comparative genomics, and experimental validation. These approaches help scientists make educated predictions about the likely functions of newly discovered proteins." }, { "id": 2310486, "postDate": "2023-06-20T11:59:00.050Z", "content": "I would strongly caution against using chatgpt or any other LLM as a reliable source of knowledge. There are several errors I spotted just from a cursory reading.
", "rawMarkdown": "I would strongly caution against using chatgpt or any other LLM as a reliable source of knowledge. There are several errors I spotted just from a cursory reading. ", "votes": 1 }, { "id": 2310823, "postDate": "2023-06-20T16:14:55.883Z", "content": "For the structure alignment tools, you could check out foldseek on this link: https://github.com/steineggerlab/foldseek
\nAnd for the detail of structure alignment, you can check out this paper: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5793808/
\nHowever, structure alignment is slow and might take serval days to run.
You can use this notebook, it is slowly but it should work on Kaggle https://www.kaggle.com/code/victorfernandezalbor/kagglefold-alphafold2-rosettafold
", "rawMarkdown": "You can use this notebook, it is slowly but it should work on Kaggle https://www.kaggle.com/code/victorfernandezalbor/kagglefold-alphafold2-rosettafold", "votes": 2 }, { "id": 2309689, "postDate": "2023-06-19T20:07:04.427Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 2309696, "author_name": "Iddo Friedberg", "author_url": "", "post_date": "2023-06-19T20:18:22.030000", "content": "Structure can inform of function, but note that AF2 predictions are also just predictions. Some proteins would have experimental X-ray or NMR structures which you may find at PDB.
\nYou can retrieve AF2 predictions for most of the superset here: https://alphafold.ebi.ac.uk/ (I'm note sure if anyone created this dataset already for this challenge?)
\nMore on prediction of function from structure https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1000485
", "votes": 1, "replies": [ { "id": 2310052, "author_name": "ADITHYA L BHAT", "author_url": "", "post_date": "2023-06-20T05:00:15.347000", "content": "assume i get the structure of the gene , how will i be able to predict its functions? I am going through the links currently .
", "votes": 0, "replies": [ { "id": 2310096, "author_name": "Evans", "author_url": "", "post_date": "2023-06-20T05:31:33.430000", "content": "You can do structure alignment search query against training protein, and use KNN to generate the confidence score
", "votes": 1, "replies": [] }, { "id": 2310339, "author_name": "ADITHYA L BHAT", "author_url": "", "post_date": "2023-06-20T09:40:17.643000", "content": "task 1 : Get the structure of the gene
\ntask 2 : structure alignment search query against training protein, and use KNN to generate the confidence score.
\ncan we have discussion about all the steps to be executed. would you like to join @victorfernandezalbor
Chat gpt answers!!!!!
\nOnce trained, AlphaFold can take a protein's sequence of amino acids and predict its shape accurately , and how do i predict its function ?
\nChatGPT
Predicting a protein's function based solely on its sequence and shape is a more complex task compared to predicting its shape alone. While protein shape provides valuable insights, understanding protein function involves considering additional factors such as evolutionary relationships, biochemical experiments, and structural analysis of protein interactions.
\nHere's a simplified explanation of how protein function prediction works:
\nSequence Analysis: The first analyze the protein\n\nHomology Modeling: a protein\n\nStructural Analysis: you mentioned, AlphaFold can predict the shape a protein accurately. Once the protein\n\nFunctional Annotations: Scientists also rely databases existing knowledge annotate assign functions proteins. These databases contain information about experimentally validated functions various proteins. comparing a newly discovered protein proteins known functions, scientists can make predictions about its potential .\n\nExperimental Validation: confirm the predicted , experimental studies are conducted. These experiments might involve laboratory techniques such enzymatic assays, protein-protein interaction assays, genetic manipulations observe the protein\nIt's important to note that predicting protein function is an ongoing area of research, and it can be challenging to accurately predict function solely based on sequence and structure. Many factors, including experimental validation, are crucial for confirming and refining these predictions.
\nOverall, predicting protein function involves a combination of sequence analysis, structural analysis, comparative genomics, and experimental validation. These approaches help scientists make educated predictions about the likely functions of newly discovered proteins.
", "votes": 0, "replies": [] }, { "id": 2310486, "author_name": "Iddo Friedberg", "author_url": "", "post_date": "2023-06-20T11:59:00.050000", "content": "I would strongly caution against using chatgpt or any other LLM as a reliable source of knowledge. There are several errors I spotted just from a cursory reading.
", "votes": 1, "replies": [] }, { "id": 2310823, "author_name": "Evans", "author_url": "", "post_date": "2023-06-20T16:14:55.883000", "content": "For the structure alignment tools, you could check out foldseek on this link: https://github.com/steineggerlab/foldseek
\nAnd for the detail of structure alignment, you can check out this paper: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5793808/
\nHowever, structure alignment is slow and might take serval days to run.
You can use this notebook, it is slowly but it should work on Kaggle https://www.kaggle.com/code/victorfernandezalbor/kagglefold-alphafold2-rosettafold
", "votes": 2, "replies": [] }, { "id": 2309689, "author_name": "", "author_url": "", "post_date": "2023-06-19T20:07:04.427000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2309535": "is it possible to use deepminds alphafold2 here ?", "2309696": "Structure can inform of function, but note that AF2 predictions are also just predictions. Some proteins would have experimental X-ray or NMR structures which you may find at [PDB](https://www.rcsb.org/).\n\nYou can retrieve AF2 predictions for most of the superset here: https://alphafold.ebi.ac.uk/ (I'm note sure if anyone created this dataset already for this challenge?)\n\nMore on prediction of function from structure https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1000485\n\n\n\n\n", "2310325": "You can use this notebook, it is slowly but it should work on Kaggle https://www.kaggle.com/code/victorfernandezalbor/kagglefold-alphafold2-rosettafold", "2309689": "" } }