Hi.
\nWe are seeing some kagglers who want to get up to speed on the bioinformatics and function prediction. The problem seems to be that there is a lot of marerial out there: where to start? I hope we can help here.
\nA good place to start is \"A (not so) quick introduction to protein function prediction\" Authored by @predragradivojac (attached). It is actually a pretty quick introduction for computational people who have zero domain knowledge and want only the relevant stuff for tacking the problems in this competition. This will explain the motivation for the problem, how it is cast, etc. The third chapter, on CAFA is a bit dated, and while historically correct, the rules that we had for previous CAFA competition do not necessarily apply to the Kaggle challenge. The only source for Rules for CAFA 5 is on this site. Also, we use slightly different assessment metrics than in section 2.3 (again, Rules for CAFA 5).
", "messages": [ { "id": 2229862, "postDate": "2023-04-21T19:11:32.963Z", "content": "Hi.
\nWe are seeing some kagglers who want to get up to speed on the bioinformatics and function prediction. The problem seems to be that there is a lot of marerial out there: where to start? I hope we can help here.
\nA good place to start is \"A (not so) quick introduction to protein function prediction\" Authored by @predragradivojac (attached). It is actually a pretty quick introduction for computational people who have zero domain knowledge and want only the relevant stuff for tacking the problems in this competition. This will explain the motivation for the problem, how it is cast, etc. The third chapter, on CAFA is a bit dated, and while historically correct, the rules that we had for previous CAFA competition do not necessarily apply to the Kaggle challenge. The only source for Rules for CAFA 5 is on this site. Also, we use slightly different assessment metrics than in section 2.3 (again, Rules for CAFA 5).
", "rawMarkdown": "Hi.\n\nWe are seeing some kagglers who want to get up to speed on the bioinformatics and function prediction. The problem seems to be that there is a lot of marerial out there: where to start? I hope we can help here. \n\nA good place to start is \"A (not so) quick introduction to protein function prediction\" Authored by @predragradivojac (attached). It is actually a pretty quick introduction for computational people who have zero domain knowledge and want only the relevant stuff for tacking the problems in this competition. This will explain the motivation for the problem, how it is cast, etc. The third chapter, on CAFA is a bit dated, and while historically correct, the rules that we had for previous CAFA competition do not necessarily apply to the Kaggle challenge. The only source for [**Rules**](https://www.kaggle.com/competitions/cafa-5-protein-function-prediction/rules) for CAFA 5 is on this site. Also, we use slightly different assessment metrics than in section 2.3 (again, Rules for CAFA 5).\n\n\n\n\n\n\n\n", "votes": 78 }, { "id": 2250839, "postDate": "2023-05-08T20:29:48.843Z", "content": "The attached PDF was way over my head. So I had ChatGPT rewrite the first section in simple terms. I have no idea if this is accurate or not…
\nProteins are important molecules in our body that perform many different tasks such as providing structure, fighting off infections, helping muscles move, and regulating processes in our cells. They make up more than half of the weight of our cells. Proteins work together with other molecules in our body, like DNA and RNA, to carry out their functions. The environment around proteins, such as other molecules and temperature, can affect how they work. Scientists study how different factors affect protein function.
\nProtein structure refers to how proteins are made up of amino acids linked together, and the different shapes they can take. Proteins can have different shapes and move around in different ways. Some proteins have a single dominant shape, while others are more flexible and can take on many shapes. The way a protein moves and changes shape is important for its function. Some proteins have parts that are ordered, while other parts are disordered and flexible. Proteins can also come together to form larger complexes, like hemoglobin.
\nWhen scientists study proteins and their functions, they often describe what the proteins do using natural language. However, in order to exchange this knowledge more easily and use computers to process large amounts of data, they often use ontologies. These are hierarchical knowledge representations, where nodes or terms describe particular activities and links represent relationships between terms. Biomedical ontologies describing protein function exist, and the Gene Ontology Consortium has developed three independent ontologies. Other ontologies involving higher levels of functional descriptors are usually species-specific. Human-centric ontologies have become sophisticated enough to be used for phenotypic annotation of proteins. Different researchers refer to protein function in different ways, and there is often confusion. Experimentalists and biocurators are both involved in annotating protein function through interpretation of experiments from the literature and summarizing these experiments using ontological representation. Biocurators are particularly important because they have deep understanding of the domain and are familiar with the ontologies.
\nData representation refers to how we express information about proteins in a way that can be understood by computers. Proteins are made up of 20 different amino acids, which can be represented as letters in an alphabet of 20 symbols. This makes it easy to use computer algorithms to compare and analyze different protein sequences. Proteins can also be represented as graphs or networks, where nodes represent proteins and links represent interactions between them. We can also look at proteins as time series data, such as measurements of gene expression at different time points. However, predicting protein function is not as simple as applying a few algorithms because proteins can have multiple forms that are encoded by the same gene and have the same amino acid sequence but different functions. This is because of events such as alternative splicing and post-translational modifications, which can produce many different protein forms with different functions.
\nInformation about proteins is kept in many biomedical databases. Every year, Nucleic Acids Research publishes a special issue that shows new databases and updates to old ones. There are also other journals that accept new biological databases. However, most of these databases are too specific and not important for this summary. There are a few that are important, which are shown in Table 1.
", "rawMarkdown": "The attached PDF was way over my head. So I had ChatGPT rewrite the first section in simple terms. I have no idea if this is accurate or not...\n\nProteins are important molecules in our body that perform many different tasks such as providing structure, fighting off infections, helping muscles move, and regulating processes in our cells. They make up more than half of the weight of our cells. Proteins work together with other molecules in our body, like DNA and RNA, to carry out their functions. The environment around proteins, such as other molecules and temperature, can affect how they work. Scientists study how different factors affect protein function.\n\nProtein structure refers to how proteins are made up of amino acids linked together, and the different shapes they can take. Proteins can have different shapes and move around in different ways. Some proteins have a single dominant shape, while others are more flexible and can take on many shapes. The way a protein moves and changes shape is important for its function. Some proteins have parts that are ordered, while other parts are disordered and flexible. Proteins can also come together to form larger complexes, like hemoglobin.\n\nWhen scientists study proteins and their functions, they often describe what the proteins do using natural language. However, in order to exchange this knowledge more easily and use computers to process large amounts of data, they often use ontologies. These are hierarchical knowledge representations, where nodes or terms describe particular activities and links represent relationships between terms. Biomedical ontologies describing protein function exist, and the Gene Ontology Consortium has developed three independent ontologies. Other ontologies involving higher levels of functional descriptors are usually species-specific. Human-centric ontologies have become sophisticated enough to be used for phenotypic annotation of proteins. Different researchers refer to protein function in different ways, and there is often confusion. Experimentalists and biocurators are both involved in annotating protein function through interpretation of experiments from the literature and summarizing these experiments using ontological representation. Biocurators are particularly important because they have deep understanding of the domain and are familiar with the ontologies.\n\nData representation refers to how we express information about proteins in a way that can be understood by computers. Proteins are made up of 20 different amino acids, which can be represented as letters in an alphabet of 20 symbols. This makes it easy to use computer algorithms to compare and analyze different protein sequences. Proteins can also be represented as graphs or networks, where nodes represent proteins and links represent interactions between them. We can also look at proteins as time series data, such as measurements of gene expression at different time points. However, predicting protein function is not as simple as applying a few algorithms because proteins can have multiple forms that are encoded by the same gene and have the same amino acid sequence but different functions. This is because of events such as alternative splicing and post-translational modifications, which can produce many different protein forms with different functions.\n\nInformation about proteins is kept in many biomedical databases. Every year, Nucleic Acids Research publishes a special issue that shows new databases and updates to old ones. There are also other journals that accept new biological databases. However, most of these databases are too specific and not important for this summary. There are a few that are important, which are shown in Table 1.", "votes": 8, "replies": [ { "id": 2342598, "postDate": "2023-07-13T02:37:03.907Z", "content": "Thanks for sharing the information, it helps a lot!!
", "rawMarkdown": "Thanks for sharing the information, it helps a lot!!" } ] }, { "id": 2267260, "postDate": "2023-05-20T18:16:23.897Z", "content": "This document is very helpful, thank you for sharing!
", "rawMarkdown": "This document is very helpful, thank you for sharing!", "votes": -1 }, { "id": 2393546, "postDate": "2023-08-16T11:23:15.237Z", "content": "Hm, is this still relevant after so many years and so much development in ML?
", "rawMarkdown": "Hm, is this still relevant after so many years and so much development in ML?" }, { "id": 2358586, "postDate": "2023-07-25T15:52:11.310Z", "content": "Thanks for sharing. It is very helpful…
", "rawMarkdown": "Thanks for sharing. It is very helpful..." }, { "id": 2246821, "postDate": "2023-05-05T13:47:39.600Z", "content": "Thanks a lot for sharing this document
", "rawMarkdown": "Thanks a lot for sharing this document \n" }, { "id": 2243265, "postDate": "2023-05-02T19:05:20.470Z", "content": "Thank you very much for sharing this document, really helpful!
", "rawMarkdown": "Thank you very much for sharing this document, really helpful!" }, { "id": 2336430, "postDate": "2023-07-09T11:26:24.657Z", "content": "The gene ontology is based on what Stephen Wolfram calls causal graph. In Wolfram's framework, it can be interpreted as a discrete spacetime.
", "rawMarkdown": "The gene ontology is based on what Stephen Wolfram calls [causal graph](https://www.wolframphysics.org/technical-introduction/the-updating-process-in-our-models/the-role-of-causal-graphs/). In Wolfram's framework, it can be interpreted as a discrete spacetime.", "votes": 1, "isDeleted": true }, { "id": 2291304, "postDate": "2023-06-07T13:10:44.510Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2271188, "postDate": "2023-05-23T17:19:15.300Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2251013, "postDate": "2023-05-09T02:12:10.923Z", "rawMarkdown": "", "isDeleted": true, "replies": [ { "id": 2251033, "postDate": "2023-05-09T02:35:56.897Z", "content": "While your intentions are good, you should get permission from the original author before putting up something like that. You are not even crediting the original author!
", "rawMarkdown": "While your intentions are good, you should get permission from the original author before putting up something like that. You are not even crediting the original author! ", "votes": 1, "replies": [ { "id": 2251037, "postDate": "2023-05-09T02:42:48.270Z", "content": "Removed, pending
", "rawMarkdown": "Removed, pending" } ] } ] }, { "id": 2344850, "postDate": "2023-07-14T19:49:26.510Z", "content": "Thanks so much for the info!
", "rawMarkdown": "Thanks so much for the info!" }, { "id": 2286432, "postDate": "2023-06-03T12:57:37.243Z", "content": "Thanks for sharing
", "rawMarkdown": "Thanks for sharing" }, { "id": 2278407, "postDate": "2023-05-28T16:49:30.570Z", "content": "Thank you for sharing !
", "rawMarkdown": "Thank you for sharing !" }, { "id": 2274866, "postDate": "2023-05-26T10:10:11.630Z", "content": "Thank you for sharing!
", "rawMarkdown": "Thank you for sharing!" }, { "id": 2272427, "postDate": "2023-05-24T13:29:06.417Z", "content": "thanks, this helps e a lot!
", "rawMarkdown": "thanks, this helps e a lot!" }, { "id": 2265166, "postDate": "2023-05-19T03:19:33.913Z", "content": "Thank you very much for sharing!
", "rawMarkdown": "Thank you very much for sharing!" }, { "id": 2248526, "postDate": "2023-05-07T00:54:39.940Z", "content": "Thanks for sharing!
", "rawMarkdown": "Thanks for sharing!" } ], "comments": [ { "id": 2250839, "author_name": "Ben Lebovitz", "author_url": "", "post_date": "2023-05-08T20:29:48.843000", "content": "The attached PDF was way over my head. So I had ChatGPT rewrite the first section in simple terms. I have no idea if this is accurate or not…
\nProteins are important molecules in our body that perform many different tasks such as providing structure, fighting off infections, helping muscles move, and regulating processes in our cells. They make up more than half of the weight of our cells. Proteins work together with other molecules in our body, like DNA and RNA, to carry out their functions. The environment around proteins, such as other molecules and temperature, can affect how they work. Scientists study how different factors affect protein function.
\nProtein structure refers to how proteins are made up of amino acids linked together, and the different shapes they can take. Proteins can have different shapes and move around in different ways. Some proteins have a single dominant shape, while others are more flexible and can take on many shapes. The way a protein moves and changes shape is important for its function. Some proteins have parts that are ordered, while other parts are disordered and flexible. Proteins can also come together to form larger complexes, like hemoglobin.
\nWhen scientists study proteins and their functions, they often describe what the proteins do using natural language. However, in order to exchange this knowledge more easily and use computers to process large amounts of data, they often use ontologies. These are hierarchical knowledge representations, where nodes or terms describe particular activities and links represent relationships between terms. Biomedical ontologies describing protein function exist, and the Gene Ontology Consortium has developed three independent ontologies. Other ontologies involving higher levels of functional descriptors are usually species-specific. Human-centric ontologies have become sophisticated enough to be used for phenotypic annotation of proteins. Different researchers refer to protein function in different ways, and there is often confusion. Experimentalists and biocurators are both involved in annotating protein function through interpretation of experiments from the literature and summarizing these experiments using ontological representation. Biocurators are particularly important because they have deep understanding of the domain and are familiar with the ontologies.
\nData representation refers to how we express information about proteins in a way that can be understood by computers. Proteins are made up of 20 different amino acids, which can be represented as letters in an alphabet of 20 symbols. This makes it easy to use computer algorithms to compare and analyze different protein sequences. Proteins can also be represented as graphs or networks, where nodes represent proteins and links represent interactions between them. We can also look at proteins as time series data, such as measurements of gene expression at different time points. However, predicting protein function is not as simple as applying a few algorithms because proteins can have multiple forms that are encoded by the same gene and have the same amino acid sequence but different functions. This is because of events such as alternative splicing and post-translational modifications, which can produce many different protein forms with different functions.
\nInformation about proteins is kept in many biomedical databases. Every year, Nucleic Acids Research publishes a special issue that shows new databases and updates to old ones. There are also other journals that accept new biological databases. However, most of these databases are too specific and not important for this summary. There are a few that are important, which are shown in Table 1.
", "votes": 8, "replies": [ { "id": 2342598, "author_name": "JUN Andy", "author_url": "", "post_date": "2023-07-13T02:37:03.907000", "content": "Thanks for sharing the information, it helps a lot!!
", "votes": 0, "replies": [] } ] }, { "id": 2267260, "author_name": "astreich", "author_url": "", "post_date": "2023-05-20T18:16:23.897000", "content": "This document is very helpful, thank you for sharing!
", "votes": -1, "replies": [] }, { "id": 2393546, "author_name": "Gyula Maloveczky4", "author_url": "", "post_date": "2023-08-16T11:23:15.237000", "content": "Hm, is this still relevant after so many years and so much development in ML?
", "votes": 0, "replies": [] }, { "id": 2358586, "author_name": "JKC", "author_url": "", "post_date": "2023-07-25T15:52:11.310000", "content": "Thanks for sharing. It is very helpful…
", "votes": 0, "replies": [] }, { "id": 2246821, "author_name": "MATHAN PRASANNA KUMAR S", "author_url": "", "post_date": "2023-05-05T13:47:39.600000", "content": "Thanks a lot for sharing this document
", "votes": 0, "replies": [] }, { "id": 2243265, "author_name": "Eli Dvoretsky", "author_url": "", "post_date": "2023-05-02T19:05:20.470000", "content": "Thank you very much for sharing this document, really helpful!
", "votes": 0, "replies": [] }, { "id": 2336430, "author_name": "", "author_url": "", "post_date": "2023-07-09T11:26:24.657000", "content": "The gene ontology is based on what Stephen Wolfram calls causal graph. In Wolfram's framework, it can be interpreted as a discrete spacetime.
", "votes": 1, "replies": [] }, { "id": 2291304, "author_name": "", "author_url": "", "post_date": "2023-06-07T13:10:44.510000", "content": "", "votes": 0, "replies": [] }, { "id": 2271188, "author_name": "", "author_url": "", "post_date": "2023-05-23T17:19:15.300000", "content": "", "votes": 0, "replies": [] }, { "id": 2251013, "author_name": "", "author_url": "", "post_date": "2023-05-09T02:12:10.923000", "content": "", "votes": 0, "replies": [ { "id": 2251033, "author_name": "Iddo Friedberg", "author_url": "", "post_date": "2023-05-09T02:35:56.897000", "content": "While your intentions are good, you should get permission from the original author before putting up something like that. You are not even crediting the original author!
", "votes": 1, "replies": [ { "id": 2251037, "author_name": "Abish Pius", "author_url": "", "post_date": "2023-05-09T02:42:48.270000", "content": "Removed, pending
", "votes": 0, "replies": [] } ] } ] }, { "id": 2344850, "author_name": "Ektor Nikolinakos", "author_url": "", "post_date": "2023-07-14T19:49:26.510000", "content": "Thanks so much for the info!
", "votes": 0, "replies": [] }, { "id": 2286432, "author_name": "L1ghtR0ast", "author_url": "", "post_date": "2023-06-03T12:57:37.243000", "content": "Thanks for sharing
", "votes": 0, "replies": [] }, { "id": 2278407, "author_name": "Julien Tetar", "author_url": "", "post_date": "2023-05-28T16:49:30.570000", "content": "Thank you for sharing !
", "votes": 0, "replies": [] }, { "id": 2274866, "author_name": "vlr", "author_url": "", "post_date": "2023-05-26T10:10:11.630000", "content": "Thank you for sharing!
", "votes": 0, "replies": [] }, { "id": 2272427, "author_name": "tanguy ", "author_url": "", "post_date": "2023-05-24T13:29:06.417000", "content": "thanks, this helps e a lot!
", "votes": 0, "replies": [] }, { "id": 2265166, "author_name": "Timofey Efimov", "author_url": "", "post_date": "2023-05-19T03:19:33.913000", "content": "Thank you very much for sharing!
", "votes": 0, "replies": [] }, { "id": 2248526, "author_name": "Autumn", "author_url": "", "post_date": "2023-05-07T00:54:39.940000", "content": "Thanks for sharing!
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2229862": "Hi.\n\nWe are seeing some kagglers who want to get up to speed on the bioinformatics and function prediction. The problem seems to be that there is a lot of marerial out there: where to start? I hope we can help here. \n\nA good place to start is \"A (not so) quick introduction to protein function prediction\" Authored by @predragradivojac (attached). It is actually a pretty quick introduction for computational people who have zero domain knowledge and want only the relevant stuff for tacking the problems in this competition. This will explain the motivation for the problem, how it is cast, etc. The third chapter, on CAFA is a bit dated, and while historically correct, the rules that we had for previous CAFA competition do not necessarily apply to the Kaggle challenge. The only source for [**Rules**](https://www.kaggle.com/competitions/cafa-5-protein-function-prediction/rules) for CAFA 5 is on this site. Also, we use slightly different assessment metrics than in section 2.3 (again, Rules for CAFA 5).\n\n\n\n\n\n\n\n", "2250839": "The attached PDF was way over my head. So I had ChatGPT rewrite the first section in simple terms. I have no idea if this is accurate or not...\n\nProteins are important molecules in our body that perform many different tasks such as providing structure, fighting off infections, helping muscles move, and regulating processes in our cells. They make up more than half of the weight of our cells. Proteins work together with other molecules in our body, like DNA and RNA, to carry out their functions. The environment around proteins, such as other molecules and temperature, can affect how they work. Scientists study how different factors affect protein function.\n\nProtein structure refers to how proteins are made up of amino acids linked together, and the different shapes they can take. Proteins can have different shapes and move around in different ways. Some proteins have a single dominant shape, while others are more flexible and can take on many shapes. The way a protein moves and changes shape is important for its function. Some proteins have parts that are ordered, while other parts are disordered and flexible. Proteins can also come together to form larger complexes, like hemoglobin.\n\nWhen scientists study proteins and their functions, they often describe what the proteins do using natural language. However, in order to exchange this knowledge more easily and use computers to process large amounts of data, they often use ontologies. These are hierarchical knowledge representations, where nodes or terms describe particular activities and links represent relationships between terms. Biomedical ontologies describing protein function exist, and the Gene Ontology Consortium has developed three independent ontologies. Other ontologies involving higher levels of functional descriptors are usually species-specific. Human-centric ontologies have become sophisticated enough to be used for phenotypic annotation of proteins. Different researchers refer to protein function in different ways, and there is often confusion. Experimentalists and biocurators are both involved in annotating protein function through interpretation of experiments from the literature and summarizing these experiments using ontological representation. Biocurators are particularly important because they have deep understanding of the domain and are familiar with the ontologies.\n\nData representation refers to how we express information about proteins in a way that can be understood by computers. Proteins are made up of 20 different amino acids, which can be represented as letters in an alphabet of 20 symbols. This makes it easy to use computer algorithms to compare and analyze different protein sequences. Proteins can also be represented as graphs or networks, where nodes represent proteins and links represent interactions between them. We can also look at proteins as time series data, such as measurements of gene expression at different time points. However, predicting protein function is not as simple as applying a few algorithms because proteins can have multiple forms that are encoded by the same gene and have the same amino acid sequence but different functions. This is because of events such as alternative splicing and post-translational modifications, which can produce many different protein forms with different functions.\n\nInformation about proteins is kept in many biomedical databases. Every year, Nucleic Acids Research publishes a special issue that shows new databases and updates to old ones. There are also other journals that accept new biological databases. However, most of these databases are too specific and not important for this summary. There are a few that are important, which are shown in Table 1.", "2267260": "This document is very helpful, thank you for sharing!", "2393546": "Hm, is this still relevant after so many years and so much development in ML?", "2358586": "Thanks for sharing. It is very helpful...", "2246821": "Thanks a lot for sharing this document \n", "2243265": "Thank you very much for sharing this document, really helpful!", "2336430": "The gene ontology is based on what Stephen Wolfram calls [causal graph](https://www.wolframphysics.org/technical-introduction/the-updating-process-in-our-models/the-role-of-causal-graphs/). In Wolfram's framework, it can be interpreted as a discrete spacetime.", "2291304": "", "2271188": "", "2251013": "", "2344850": "Thanks so much for the info!", "2286432": "Thanks for sharing", "2278407": "Thank you for sharing !", "2274866": "Thank you for sharing!", "2272427": "thanks, this helps e a lot!", "2265166": "Thank you very much for sharing!", "2248526": "Thanks for sharing!" } }