I have no idea on how to solve the whole problem, but just looking at the RNA string sequences, I'm wondering about using an algorithm called K-medoids to at least get some minimal, useful data-mining by clustering the RNA-sequence strings. Basically, you could use K-medoids (a discrete version of K-means that works in arbitrary metric spaces), and then use the \"edit-distance\" metric: the edit-distance dist(A, B) = # of minimum single-character changes to transform string A into string B, and vice versa. Of course this doesn't solve the prediction problem, but I think this could be a relatively easy way to get some minimum data mining out of the RNA sequences.
\nI believe the \"pyclustering library\" (https://pyclustering.github.io/) has a section on the k-medoids algorithm, and the edit-distance metric could be programmed as a function in Python to use in the clustering algorithm.
", "messages": [ { "id": "2440948", "postDate": "09/15/2023 20:39:41", "content": "I have no idea on how to solve the whole problem, but just looking at the RNA string sequences, I'm wondering about using an algorithm called K-medoids to at least get some minimal, useful data-mining by clustering the RNA-sequence strings. Basically, you could use K-medoids (a discrete version of K-means that works in arbitrary metric spaces), and then use the \"edit-distance\" metric: the edit-distance dist(A, B) = # of minimum single-character changes to transform string A into string B, and vice versa. Of course this doesn't solve the prediction problem, but I think this could be a relatively easy way to get some minimum data mining out of the RNA sequences.
\nI believe the \"pyclustering library\" (https://pyclustering.github.io/) has a section on the k-medoids algorithm, and the edit-distance metric could be programmed as a function in Python to use in the clustering algorithm.
", "rawMarkdown": "I have no idea on how to solve the whole problem, but just looking at the RNA string sequences, I'm wondering about using an algorithm called K-medoids to at least get some minimal, useful data-mining by clustering the RNA-sequence strings. Basically, you could use K-medoids (a discrete version of K-means that works in arbitrary metric spaces), and then use the \"edit-distance\" metric: the edit-distance dist(A, B) = # of minimum single-character changes to transform string A into string B, and vice versa. Of course this doesn't solve the prediction problem, but I think this could be a relatively easy way to get some minimum data mining out of the RNA sequences.\n\nI believe the \"pyclustering library\" (https://pyclustering.github.io/) has a section on the k-medoids algorithm, and the edit-distance metric could be programmed as a function in Python to use in the clustering algorithm.", "votes": null }, { "id": "2442692", "postDate": "09/17/2023 07:12:22", "content": "Keep working on it….
", "rawMarkdown": "Keep working on it....", "votes": null }, { "id": "2444740", "postDate": "09/18/2023 12:59:18", "content": "Look at: https://en.wikipedia.org/wiki/Sequence_clustering
\nCH-hit: https://www.kaggle.com/code/alexandervc/cd-hit-sequence-clustering
\nDiamond + graph clustering: https://www.kaggle.com/code/alexandervc/cafa5-23-groups-and-folds-diamond-igraph
Thanks @Alexander! I'll check these resources out.
", "rawMarkdown": "Thanks @Alexander! I'll check these resources out.", "votes": null } ], "comments": [ { "id": 2442692, "author_name": "josealexanderrubio", "author_url": "", "post_date": "09/17/2023 07:12:22", "content": "Keep working on it….
", "votes": null, "replies": [] }, { "id": 2444740, "author_name": "alexandervc", "author_url": "", "post_date": "09/18/2023 12:59:18", "content": "Look at: https://en.wikipedia.org/wiki/Sequence_clustering
\nCH-hit: https://www.kaggle.com/code/alexandervc/cd-hit-sequence-clustering
\nDiamond + graph clustering: https://www.kaggle.com/code/alexandervc/cafa5-23-groups-and-folds-diamond-igraph
Thanks @Alexander! I'll check these resources out.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2440948": "I have no idea on how to solve the whole problem, but just looking at the RNA string sequences, I'm wondering about using an algorithm called K-medoids to at least get some minimal, useful data-mining by clustering the RNA-sequence strings. Basically, you could use K-medoids (a discrete version of K-means that works in arbitrary metric spaces), and then use the \"edit-distance\" metric: the edit-distance dist(A, B) = # of minimum single-character changes to transform string A into string B, and vice versa. Of course this doesn't solve the prediction problem, but I think this could be a relatively easy way to get some minimum data mining out of the RNA sequences.\n\nI believe the \"pyclustering library\" (https://pyclustering.github.io/) has a section on the k-medoids algorithm, and the edit-distance metric could be programmed as a function in Python to use in the clustering algorithm.", "2442692": "Keep working on it....", "2444740": "Look at: https://en.wikipedia.org/wiki/Sequence_clustering\nCH-hit: https://www.kaggle.com/code/alexandervc/cd-hit-sequence-clustering\nDiamond + graph clustering: https://www.kaggle.com/code/alexandervc/cafa5-23-groups-and-folds-diamond-igraph", "2445448": "Thanks @Alexander! I'll check these resources out." }, "source": "meta" }