![](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F779570%2F5272afd5d8525d1dc52993c9834828a5%2FUntitled.jpg?generation=1714258174972065&alt=media\")
Based on the literature, to my read there are established \"known\" binding pockets for many proteins. And there could also be novel (undiscovered) binding pockets as well. Still, I'd like to understand whether the known binding pockets are pretty easy to find and understand, if you know what you're doing?
\nIs it super simple? Papers and datasets seem to refer to protein/ligand known combos, and thus if you know the structure, and which is the protein and which the ligand, the binding pocket is obviously where they connect? But I also don't have any idea where protein plus bound ligand PDBs live?
\nSo one basic search example would be something like this for sEH (link from data page). On the sequence page, there's:
\nAnd I'm wondering if any or all of these are important, and for a given row like BINDING SITE (34N), if important, are all the notated sites all binding pockets, only some, or…?
\nAnd that's just one example since I don't know what I'm looking for, but wanted to add specifics. Is the pdb file itself a better way to discover notated binding pockets, or the UniProt website, or…?
", "messages": [ { "id": "2779826", "postDate": "04/27/2024 20:50:09", "content": "Based on the literature, to my read there are established \"known\" binding pockets for many proteins. And there could also be novel (undiscovered) binding pockets as well. Still, I'd like to understand whether the known binding pockets are pretty easy to find and understand, if you know what you're doing?
\nIs it super simple? Papers and datasets seem to refer to protein/ligand known combos, and thus if you know the structure, and which is the protein and which the ligand, the binding pocket is obviously where they connect? But I also don't have any idea where protein plus bound ligand PDBs live?
\nSo one basic search example would be something like this for sEH (link from data page). On the sequence page, there's:
\nAnd I'm wondering if any or all of these are important, and for a given row like BINDING SITE (34N), if important, are all the notated sites all binding pockets, only some, or…?
\nAnd that's just one example since I don't know what I'm looking for, but wanted to add specifics. Is the pdb file itself a better way to discover notated binding pockets, or the UniProt website, or…?
", "rawMarkdown": "Based on the literature, to my read there are established \"known\" binding pockets for many proteins. And there could also be novel (undiscovered) binding pockets as well. Still, I'd like to understand whether the known binding pockets are pretty easy to find and understand, if you know what you're doing?\n\nIs it super simple? Papers and datasets seem to refer to protein/ligand known combos, and thus if you know the structure, and which is the protein and which the ligand, the binding pocket is obviously where they connect? But I also don't have any idea where protein plus bound ligand PDBs live?\n\nSo one basic search example would be something like [this for sEH](https://www.rcsb.org/sequence/3I28) (link from data page). On the sequence page, there's:\n* BINDING SITE\n* BINDING SITE (34N)\n* ACTIVE SITE\n\nAnd I'm wondering if any or all of these are important, and for a given row like BINDING SITE (34N), if important, are all the notated sites all binding pockets, only some, or...?\n\nAnd that's just one example since I don't know what I'm looking for, but wanted to add specifics. Is the pdb file itself a better way to discover notated binding pockets, or the UniProt website, or...?", "votes": null }, { "id": "2779888", "postDate": "04/27/2024 21:45:36", "content": "This is a little out of my expertise tbh, so I'd be keen to learn more here too.
\nRegarding protein+ligand bound PDBs: each PDB entry is a specific structure. These can be apo structures (no ligand bound) or they can contain ligand. Below I've searched specifically for sEH structures.
\nExample: Here's a few PDB structures for sEH with different ligands bound Example
\nRegarding binding site vs active site: the active site is often where an enzyme catalyzes a chemical reaction. Active sites are a subset of binding sites.
\nHope this helps. Looking forward to learning more from other replies.
", "rawMarkdown": "This is a little out of my expertise tbh, so I'd be keen to learn more here too.\n\nRegarding protein+ligand bound PDBs: each PDB entry is a specific structure. These can be apo structures (no ligand bound) or they can contain ligand. Below I've searched specifically for sEH structures.\n\nExample: Here's a few PDB structures for sEH with different ligands bound [Example](https://www.rcsb.org/search?request=%7B%22query%22%3A%7B%22type%22%3A%22group%22%2C%22nodes%22%3A%5B%7B%22type%22%3A%22group%22%2C%22nodes%22%3A%5B%7B%22type%22%3A%22group%22%2C%22nodes%22%3A%5B%7B%22type%22%3A%22terminal%22%2C%22service%22%3A%22text%22%2C%22parameters%22%3A%7B%22attribute%22%3A%22struct_keywords.text%22%2C%22operator%22%3A%22contains_phrase%22%2C%22value%22%3A%22Inhibitor%2C%20Complex%2C%20sEH%2C%20Hydrolase%22%7D%7D%5D%2C%22logical_operator%22%3A%22and%22%7D%5D%2C%22logical_operator%22%3A%22and%22%2C%22label%22%3A%22text%22%7D%5D%2C%22logical_operator%22%3A%22and%22%7D%2C%22return_type%22%3A%22entry%22%2C%22request_options%22%3A%7B%22paginate%22%3A%7B%22start%22%3A0%2C%22rows%22%3A25%7D%2C%22results_content_type%22%3A%5B%22experimental%22%5D%2C%22sort%22%3A%5B%7B%22sort_by%22%3A%22score%22%2C%22direction%22%3A%22desc%22%7D%5D%2C%22scoring_strategy%22%3A%22combined%22%7D%2C%22request_info%22%3A%7B%22query_id%22%3A%224fa95e1c96fa197c75da891854a0e844%22%7D%7D)\n\nRegarding binding site vs active site: the active site is often where an enzyme catalyzes a chemical reaction. Active sites are a subset of binding sites.\n\nHope this helps. Looking forward to learning more from other replies.", "votes": null }, { "id": "2779923", "postDate": "04/27/2024 23:08:39", "content": "Take a look at the dataset I created here - LeashBio - BELKA - Proteins. It has a total of 270 structures for the 3 protein of interest. To view them, I recommend using PyMOL, but any other viewer would work, and you can view individual files on RCSB-PDB by clicking on the Structure link at the top and at the left of the page.
\nAs for where an inhibitor binds, that's not always obvious from a structure like the competition hosts mentioned for sEH, 3i28, because there is no ligand bound in that structure - it is an \"apo-\"structure. But, luckily, there are lots of other structures for sEH available and many of them contain bound ligands, for example, 1zd2. Here's a picture I made in PyMOL of that structure with the ligand shown with carbons colored cyan. Notice that the left, phosphatase lobe also has something bound. In this case, I believe that is a free phosphate group from the solution, but I have looked at it closely. There are other structures that have other small molecules bound in that site, but I haven't paid attention to them.
\n
Notice here that the protein (magenta) has 2 lobes. The one on the left in the picture has phosphatase activity (removes phosphates from other proteins) and the one on the right has epoxide hydrolase activity. The lobe on the right is the one we're interested in here, so the location of the highlighted ligand shows us a putative binding site. (I learned about the different lobes reading some short background pages on sEH.) If we look at many structures of sEH, we can see a common theme emerge. Here's a picture of the ligands (if one is present) from the 91 sEH structures in my dataset.
\n![](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F779570%2F6e38385e1e4172c5fbf27344de7c5cd1%2Fallligands.jpg?generation=1714258892573566&alt=media\")
This gives us very high confidence of where the binding site is. Here's a zoomed in picture of the binding site of the protein with a ligand illustrated. It is really more like a binding channel in this case, but I'm not 100% sure how the enzyme works with respect to binding.
\n![](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F779570%2Feb23b5b5f97e14f6495e7f0bd8c366d3%2Fzoomed.jpg?generation=1714259194348620&alt=media\")
So, after this probably too long post, it all comes down to some domain knowledge and being familiar with looking at 3D structures of proteins. I can't recommend enough looking at the structures with the 3D viewer on the PDB site or PyMOL. Once you get familiar with looking at 3D structures, you'll think about this stuff in a completely new way.
", "rawMarkdown": "Take a look at the dataset I created here - [LeashBio - BELKA - Proteins](https://www.kaggle.com/datasets/kirkdco/leashbio-belka-proteins). It has a total of 270 structures for the 3 protein of interest. To view them, I recommend using [PyMOL](https://pymol.org/), but any other viewer would work, and you can view individual files on RCSB-PDB by clicking on the Structure link at the top and at the left of the page.\n\nAs for where an inhibitor binds, that's not always obvious from a structure like the competition hosts mentioned for sEH, [3i28](https://www.rcsb.org/structure/3i28), because there is no ligand bound in that structure - it is an \"apo-\"structure. But, luckily, there are lots of other structures for sEH available and many of them contain bound ligands, for example, [1zd2](https://www.rcsb.org/structure/1ZD2). Here's a picture I made in PyMOL of that structure with the ligand shown with carbons colored cyan. Notice that the left, phosphatase lobe also has something bound. In this case, I believe that is a free phosphate group from the solution, but I have looked at it closely. There are other structures that have other small molecules bound in that site, but I haven't paid attention to them.\n\n\n\nNotice here that the protein (magenta) has 2 lobes. The one on the left in the picture has phosphatase activity (removes phosphates from other proteins) and the one on the right has epoxide hydrolase activity. The lobe on the right is the one we're interested in here, so the location of the highlighted ligand shows us a putative binding site. (I learned about the different lobes reading some short background pages on sEH.) If we look at many structures of sEH, we can see a common theme emerge. Here's a picture of the ligands (if one is present) from the 91 sEH structures in my dataset. \n\n\n\nThis gives us very high confidence of where the binding site is. Here's a zoomed in picture of the binding site of the protein with a ligand illustrated. It is really more like a binding channel in this case, but I'm not 100% sure how the enzyme works with respect to binding. \n\n\n\nSo, after this probably too long post, it all comes down to some domain knowledge and being familiar with looking at 3D structures of proteins. I can't recommend enough looking at the structures with the 3D viewer on the PDB site or PyMOL. Once you get familiar with looking at 3D structures, you'll think about this stuff in a completely new way.", "votes": null }, { "id": "2779941", "postDate": "04/28/2024 00:00:34", "content": "\n\nHere's a picture of the ligands (if one is present) from the 91 sEH structures in my dataset.
\n
So that's showing a whole bunch overlaid on top of each other, showing that they all bind in the same spot? (or at least near to the same)
\nI'll take a look and add if I have more questions after trying to use it… :)
", "rawMarkdown": "> Here's a picture of the ligands (if one is present) from the 91 sEH structures in my dataset.\n\nSo that's showing a whole bunch overlaid on top of each other, showing that they all bind in the same spot? (or at least near to the same)\n\nI'll take a look and add if I have more questions after trying to use it... :)", "votes": null }, { "id": "2780019", "postDate": "04/28/2024 01:41:46", "content": "Yes, that's exactly right. Sorry, I left out a step in the description.
\nThe 91 structures I have for sEH all have the same protein and I aligned their structures based on the protein's backbone carbons. Everything else in the structure comes along for the ride. The ligands picture I showed has all the ligands shown but I turned off all but one protein so the ligands are easier to see. They all pile up on top of each other in the binding pocket, and since many of them are labelled as \"inhibitors\" of sEH, this seems like an important spot. 😉
\nYes, please ask away! Happy to answer questions if I can.
", "rawMarkdown": "Yes, that's exactly right. Sorry, I left out a step in the description.\n\nThe 91 structures I have for sEH all have the same protein and I aligned their structures based on the protein's backbone carbons. Everything else in the structure comes along for the ride. The ligands picture I showed has all the ligands shown but I turned off all but one protein so the ligands are easier to see. They all pile up on top of each other in the binding pocket, and since many of them are labelled as \"inhibitors\" of sEH, this seems like an important spot. 😉\n\nYes, please ask away! Happy to answer questions if I can.", "votes": null } ], "comments": [ { "id": 2779888, "author_name": "chemdatafarmer", "author_url": "", "post_date": "04/27/2024 21:45:36", "content": "This is a little out of my expertise tbh, so I'd be keen to learn more here too.
\nRegarding protein+ligand bound PDBs: each PDB entry is a specific structure. These can be apo structures (no ligand bound) or they can contain ligand. Below I've searched specifically for sEH structures.
\nExample: Here's a few PDB structures for sEH with different ligands bound Example
\nRegarding binding site vs active site: the active site is often where an enzyme catalyzes a chemical reaction. Active sites are a subset of binding sites.
\nHope this helps. Looking forward to learning more from other replies.
", "votes": null, "replies": [] }, { "id": 2779923, "author_name": "kirkdco", "author_url": "", "post_date": "04/27/2024 23:08:39", "content": "Take a look at the dataset I created here - LeashBio - BELKA - Proteins. It has a total of 270 structures for the 3 protein of interest. To view them, I recommend using PyMOL, but any other viewer would work, and you can view individual files on RCSB-PDB by clicking on the Structure link at the top and at the left of the page.
\nAs for where an inhibitor binds, that's not always obvious from a structure like the competition hosts mentioned for sEH, 3i28, because there is no ligand bound in that structure - it is an \"apo-\"structure. But, luckily, there are lots of other structures for sEH available and many of them contain bound ligands, for example, 1zd2. Here's a picture I made in PyMOL of that structure with the ligand shown with carbons colored cyan. Notice that the left, phosphatase lobe also has something bound. In this case, I believe that is a free phosphate group from the solution, but I have looked at it closely. There are other structures that have other small molecules bound in that site, but I haven't paid attention to them.
\n
Notice here that the protein (magenta) has 2 lobes. The one on the left in the picture has phosphatase activity (removes phosphates from other proteins) and the one on the right has epoxide hydrolase activity. The lobe on the right is the one we're interested in here, so the location of the highlighted ligand shows us a putative binding site. (I learned about the different lobes reading some short background pages on sEH.) If we look at many structures of sEH, we can see a common theme emerge. Here's a picture of the ligands (if one is present) from the 91 sEH structures in my dataset.
\n
This gives us very high confidence of where the binding site is. Here's a zoomed in picture of the binding site of the protein with a ligand illustrated. It is really more like a binding channel in this case, but I'm not 100% sure how the enzyme works with respect to binding.
\n
So, after this probably too long post, it all comes down to some domain knowledge and being familiar with looking at 3D structures of proteins. I can't recommend enough looking at the structures with the 3D viewer on the PDB site or PyMOL. Once you get familiar with looking at 3D structures, you'll think about this stuff in a completely new way.
", "votes": null, "replies": [ { "id": 2779941, "author_name": "roberthatch", "author_url": "", "post_date": "04/28/2024 00:00:34", "content": "\n\nHere's a picture of the ligands (if one is present) from the 91 sEH structures in my dataset.
\n
So that's showing a whole bunch overlaid on top of each other, showing that they all bind in the same spot? (or at least near to the same)
\nI'll take a look and add if I have more questions after trying to use it… :)
", "votes": null, "replies": [ { "id": 2780019, "author_name": "kirkdco", "author_url": "", "post_date": "04/28/2024 01:41:46", "content": "Yes, that's exactly right. Sorry, I left out a step in the description.
\nThe 91 structures I have for sEH all have the same protein and I aligned their structures based on the protein's backbone carbons. Everything else in the structure comes along for the ride. The ligands picture I showed has all the ligands shown but I turned off all but one protein so the ligands are easier to see. They all pile up on top of each other in the binding pocket, and since many of them are labelled as \"inhibitors\" of sEH, this seems like an important spot. 😉
\nYes, please ask away! Happy to answer questions if I can.
", "votes": null, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "2779826": "Based on the literature, to my read there are established \"known\" binding pockets for many proteins. And there could also be novel (undiscovered) binding pockets as well. Still, I'd like to understand whether the known binding pockets are pretty easy to find and understand, if you know what you're doing?\n\nIs it super simple? Papers and datasets seem to refer to protein/ligand known combos, and thus if you know the structure, and which is the protein and which the ligand, the binding pocket is obviously where they connect? But I also don't have any idea where protein plus bound ligand PDBs live?\n\nSo one basic search example would be something like [this for sEH](https://www.rcsb.org/sequence/3I28) (link from data page). On the sequence page, there's:\n* BINDING SITE\n* BINDING SITE (34N)\n* ACTIVE SITE\n\nAnd I'm wondering if any or all of these are important, and for a given row like BINDING SITE (34N), if important, are all the notated sites all binding pockets, only some, or...?\n\nAnd that's just one example since I don't know what I'm looking for, but wanted to add specifics. Is the pdb file itself a better way to discover notated binding pockets, or the UniProt website, or...?", "2779888": "This is a little out of my expertise tbh, so I'd be keen to learn more here too.\n\nRegarding protein+ligand bound PDBs: each PDB entry is a specific structure. These can be apo structures (no ligand bound) or they can contain ligand. Below I've searched specifically for sEH structures.\n\nExample: Here's a few PDB structures for sEH with different ligands bound [Example](https://www.rcsb.org/search?request=%7B%22query%22%3A%7B%22type%22%3A%22group%22%2C%22nodes%22%3A%5B%7B%22type%22%3A%22group%22%2C%22nodes%22%3A%5B%7B%22type%22%3A%22group%22%2C%22nodes%22%3A%5B%7B%22type%22%3A%22terminal%22%2C%22service%22%3A%22text%22%2C%22parameters%22%3A%7B%22attribute%22%3A%22struct_keywords.text%22%2C%22operator%22%3A%22contains_phrase%22%2C%22value%22%3A%22Inhibitor%2C%20Complex%2C%20sEH%2C%20Hydrolase%22%7D%7D%5D%2C%22logical_operator%22%3A%22and%22%7D%5D%2C%22logical_operator%22%3A%22and%22%2C%22label%22%3A%22text%22%7D%5D%2C%22logical_operator%22%3A%22and%22%7D%2C%22return_type%22%3A%22entry%22%2C%22request_options%22%3A%7B%22paginate%22%3A%7B%22start%22%3A0%2C%22rows%22%3A25%7D%2C%22results_content_type%22%3A%5B%22experimental%22%5D%2C%22sort%22%3A%5B%7B%22sort_by%22%3A%22score%22%2C%22direction%22%3A%22desc%22%7D%5D%2C%22scoring_strategy%22%3A%22combined%22%7D%2C%22request_info%22%3A%7B%22query_id%22%3A%224fa95e1c96fa197c75da891854a0e844%22%7D%7D)\n\nRegarding binding site vs active site: the active site is often where an enzyme catalyzes a chemical reaction. Active sites are a subset of binding sites.\n\nHope this helps. Looking forward to learning more from other replies.", "2779923": "Take a look at the dataset I created here - [LeashBio - BELKA - Proteins](https://www.kaggle.com/datasets/kirkdco/leashbio-belka-proteins). It has a total of 270 structures for the 3 protein of interest. To view them, I recommend using [PyMOL](https://pymol.org/), but any other viewer would work, and you can view individual files on RCSB-PDB by clicking on the Structure link at the top and at the left of the page.\n\nAs for where an inhibitor binds, that's not always obvious from a structure like the competition hosts mentioned for sEH, [3i28](https://www.rcsb.org/structure/3i28), because there is no ligand bound in that structure - it is an \"apo-\"structure. But, luckily, there are lots of other structures for sEH available and many of them contain bound ligands, for example, [1zd2](https://www.rcsb.org/structure/1ZD2). Here's a picture I made in PyMOL of that structure with the ligand shown with carbons colored cyan. Notice that the left, phosphatase lobe also has something bound. In this case, I believe that is a free phosphate group from the solution, but I have looked at it closely. There are other structures that have other small molecules bound in that site, but I haven't paid attention to them.\n\n\n\nNotice here that the protein (magenta) has 2 lobes. The one on the left in the picture has phosphatase activity (removes phosphates from other proteins) and the one on the right has epoxide hydrolase activity. The lobe on the right is the one we're interested in here, so the location of the highlighted ligand shows us a putative binding site. (I learned about the different lobes reading some short background pages on sEH.) If we look at many structures of sEH, we can see a common theme emerge. Here's a picture of the ligands (if one is present) from the 91 sEH structures in my dataset. \n\n\n\nThis gives us very high confidence of where the binding site is. Here's a zoomed in picture of the binding site of the protein with a ligand illustrated. It is really more like a binding channel in this case, but I'm not 100% sure how the enzyme works with respect to binding. \n\n\n\nSo, after this probably too long post, it all comes down to some domain knowledge and being familiar with looking at 3D structures of proteins. I can't recommend enough looking at the structures with the 3D viewer on the PDB site or PyMOL. Once you get familiar with looking at 3D structures, you'll think about this stuff in a completely new way.", "2779941": "> Here's a picture of the ligands (if one is present) from the 91 sEH structures in my dataset.\n\nSo that's showing a whole bunch overlaid on top of each other, showing that they all bind in the same spot? (or at least near to the same)\n\nI'll take a look and add if I have more questions after trying to use it... :)", "2780019": "Yes, that's exactly right. Sorry, I left out a step in the description.\n\nThe 91 structures I have for sEH all have the same protein and I aligned their structures based on the protein's backbone carbons. Everything else in the structure comes along for the ride. The ligands picture I showed has all the ligands shown but I turned off all but one protein so the ligands are easier to see. They all pile up on top of each other in the binding pocket, and since many of them are labelled as \"inhibitors\" of sEH, this seems like an important spot. 😉\n\nYes, please ask away! Happy to answer questions if I can." }, "source": "meta" }