Hi everyone
\nafter a few try I notice that if you graph the RNA sequence on a 3D plane you will notice that it doesn't look like a helic shape. what is going on here ? some does a bit but not good, quite a lot of them has a random shape. am I wrong for feeling like these data quite random ?
", "messages": [ { "id": "3144822", "postDate": "03/08/2025 22:41:47", "content": "Hi everyone
\nafter a few try I notice that if you graph the RNA sequence on a 3D plane you will notice that it doesn't look like a helic shape. what is going on here ? some does a bit but not good, quite a lot of them has a random shape. am I wrong for feeling like these data quite random ?
", "rawMarkdown": "Hi everyone\n\nafter a few try I notice that if you graph the RNA sequence on a 3D plane you will notice that it doesn't look like a helic shape. what is going on here ? some does a bit but not good, quite a lot of them has a random shape. am I wrong for feeling like these data quite random ?", "votes": null }, { "id": "3145027", "postDate": "03/09/2025 08:33:38", "content": "RNA is not DNA.
\nDNA has a double helix structure
\nRNA is less structured, similar to proteins which can fold in more diverse structures
A straightforward explanation is that the coordinates you’re seeing in the dataset are for the C1′ atom of each nucleotide, not for the full backbone or base pairs. Typically, one expects an “RNA helix” shape if you plot the entire phosphate–sugar backbone or when the bases are paired in a canonical double helix. In these files, we only have a single point per residue (the C1′ atom), so the resulting path can look quite irregular, especially for larger loops or bulges in the RNA structure. In addition, many of these RNAs are not strictly helical but have complex junctions and 3D motifs—so a random-looking 3D path is normal.
", "rawMarkdown": "A straightforward explanation is that the coordinates you’re seeing in the dataset are for the C1′ atom of each nucleotide, not for the full backbone or base pairs. Typically, one expects an “RNA helix” shape if you plot the entire phosphate–sugar backbone or when the bases are paired in a canonical double helix. In these files, we only have a single point per residue (the C1′ atom), so the resulting path can look quite irregular, especially for larger loops or bulges in the RNA structure. In addition, many of these RNAs are not strictly helical but have complex junctions and 3D motifs—so a random-looking 3D path is normal.", "votes": null }, { "id": "3148538", "postDate": "03/13/2025 09:00:35", "content": "It turns out that the task is more complicated. Perhaps the base centers would be easier to predict?
", "rawMarkdown": "It turns out that the task is more complicated. Perhaps the base centers would be easier to predict?", "votes": null }, { "id": "3148539", "postDate": "03/13/2025 09:04:59", "content": "You’re right that, by focusing on only the C1′ atoms, we get a single point per nucleotide—and that alone doesn’t always capture the familiar helix geometry or give an intuitive sense of base–base interactions. In principle, using base centers or other representative atoms might produce more recognizable geometry.
", "rawMarkdown": "You’re right that, by focusing on only the C1′ atoms, we get a single point per nucleotide—and that alone doesn’t always capture the familiar helix geometry or give an intuitive sense of base–base interactions. In principle, using base centers or other representative atoms might produce more recognizable geometry.", "votes": null }, { "id": "3152063", "postDate": "03/17/2025 11:30:00", "content": "so will using data from other successful model will help? like Alphafold, trRosetteRNA etc. I have to extract them them specifically C1'
", "rawMarkdown": "so will using data from other successful model will help? like Alphafold, trRosetteRNA etc. I have to extract them them specifically C1'", "votes": null }, { "id": "3152069", "postDate": "03/17/2025 11:40:05", "content": "I think leveraging richer atomic coordinates (e.g., full backbone or base centers from AlphaFold/trRosettaRNA) can improve structural representation, though it increases complexity. No harm in testing it out as it is the core of the competition.
", "rawMarkdown": "I think leveraging richer atomic coordinates (e.g., full backbone or base centers from AlphaFold/trRosettaRNA) can improve structural representation, though it increases complexity. No harm in testing it out as it is the core of the competition.", "votes": null } ], "comments": [ { "id": 3145027, "author_name": "louisstefanuto", "author_url": "", "post_date": "03/09/2025 08:33:38", "content": "RNA is not DNA.
\nDNA has a double helix structure
\nRNA is less structured, similar to proteins which can fold in more diverse structures
A straightforward explanation is that the coordinates you’re seeing in the dataset are for the C1′ atom of each nucleotide, not for the full backbone or base pairs. Typically, one expects an “RNA helix” shape if you plot the entire phosphate–sugar backbone or when the bases are paired in a canonical double helix. In these files, we only have a single point per residue (the C1′ atom), so the resulting path can look quite irregular, especially for larger loops or bulges in the RNA structure. In addition, many of these RNAs are not strictly helical but have complex junctions and 3D motifs—so a random-looking 3D path is normal.
", "votes": null, "replies": [ { "id": 3148538, "author_name": "sapr3s", "author_url": "", "post_date": "03/13/2025 09:00:35", "content": "It turns out that the task is more complicated. Perhaps the base centers would be easier to predict?
", "votes": null, "replies": [ { "id": 3148539, "author_name": "younusmohamed", "author_url": "", "post_date": "03/13/2025 09:04:59", "content": "You’re right that, by focusing on only the C1′ atoms, we get a single point per nucleotide—and that alone doesn’t always capture the familiar helix geometry or give an intuitive sense of base–base interactions. In principle, using base centers or other representative atoms might produce more recognizable geometry.
", "votes": null, "replies": [ { "id": 3152063, "author_name": "leecharlette", "author_url": "", "post_date": "03/17/2025 11:30:00", "content": "so will using data from other successful model will help? like Alphafold, trRosetteRNA etc. I have to extract them them specifically C1'
", "votes": null, "replies": [ { "id": 3152069, "author_name": "younusmohamed", "author_url": "", "post_date": "03/17/2025 11:40:05", "content": "I think leveraging richer atomic coordinates (e.g., full backbone or base centers from AlphaFold/trRosettaRNA) can improve structural representation, though it increases complexity. No harm in testing it out as it is the core of the competition.
", "votes": null, "replies": [] } ] } ] } ] } ] } ], "raw_markdown_by_id": { "3144822": "Hi everyone\n\nafter a few try I notice that if you graph the RNA sequence on a 3D plane you will notice that it doesn't look like a helic shape. what is going on here ? some does a bit but not good, quite a lot of them has a random shape. am I wrong for feeling like these data quite random ?", "3145027": "RNA is not DNA.\nDNA has a double helix structure\nRNA is less structured, similar to proteins which can fold in more diverse structures", "3148530": "A straightforward explanation is that the coordinates you’re seeing in the dataset are for the C1′ atom of each nucleotide, not for the full backbone or base pairs. Typically, one expects an “RNA helix” shape if you plot the entire phosphate–sugar backbone or when the bases are paired in a canonical double helix. In these files, we only have a single point per residue (the C1′ atom), so the resulting path can look quite irregular, especially for larger loops or bulges in the RNA structure. In addition, many of these RNAs are not strictly helical but have complex junctions and 3D motifs—so a random-looking 3D path is normal.", "3148538": "It turns out that the task is more complicated. Perhaps the base centers would be easier to predict?", "3148539": "You’re right that, by focusing on only the C1′ atoms, we get a single point per nucleotide—and that alone doesn’t always capture the familiar helix geometry or give an intuitive sense of base–base interactions. In principle, using base centers or other representative atoms might produce more recognizable geometry.", "3152063": "so will using data from other successful model will help? like Alphafold, trRosetteRNA etc. I have to extract them them specifically C1'", "3152069": "I think leveraging richer atomic coordinates (e.g., full backbone or base centers from AlphaFold/trRosettaRNA) can improve structural representation, though it increases complexity. No harm in testing it out as it is the core of the competition." }, "source": "meta" }