Dr. Jensen, which is better between the Tanimoto similarities and the DICE similarities? for example I will do molecular docking, First I have to match the test compound (test ligand) to the natural compound (native ligand) of the protein The 2nd test ligand with the best similarity index will be carried out by docking the protein molecule I'm glad you will upload a video explanation to compare Tanimoto Vs DICE using rDKit,
Hello Dr. Jensen, may I ask what is the basis of the nBits=1024 and is it possible to expand this function so I can compare the bitvectors of all indices in the tuple with each other (like you compare index zero with the rest then move to index 1 with the rest and so on)? I have over 500 compounds in my dataset that needed to be clustered and the molecular weight is 1kDa (I am hoping to have a full picture of compounds with similar binding efficiency to a receptor pocket. Thank you very much.
If I understand your question correctly, you want to calculate the Tanimoto similarity between all pairs of molecules. That's certainly possible, just make a double for loop for all your molecules. If you get stuck, consider using the RDKit mailing list: sourceforge.net/p/rdkit/mailman/rdkit-discuss/ . There's a lot of smart people there who are usually happy to help.
I also subscribed to the rdkit mailing list. I watched your video because the RDKit documentation is loaded with too much info and I would like to commend you for making a very simple and easy to understand explanation of how to perform Tanimoto similarity because I also plan to use the same. I just asked if it is possible to perform this on large data sets that will scan the full fingerprint like what you did to Cysteine and Glycine. May I know what is the basis for 1024 number of bits?
Hello Dr. Jensen, May I ask how exactly "GetMorganFingerprint" tell us? From what I only know is that any non-zero value indicates that there is a neighbor around, what does "0" tell us, and how to know which atom is the center one?
The fingerprint code makes fragments of various sizes and centered on every atom in the molecule. A "0" and a "1" indicate the absence and presence of a particular fragment in the molecule
It's not strictly possible to convert a fingerprint to a SMILES string. Some people have trained machine learning models to to this, but it's still experimental.
Jan Jensen Sir i am new to ml I have dataset , in which molecule is hashed in to 1024 bit could you please give me suggestions to build a model to predict weather the molecule is toxic or non toxic Thank you
Hey I am a student(engineering ) from India and I am currently working a project for helping the school students for conducting their chemistry exam through a platform so i need more information about the methods that can be used to verify the structures drawn by the students. So I need your support can I personally contact u or plz provide me your email address for asking some doubts .
That was a very useful video. It's my first time ever using RDKit, but this video and the previous one made everything very clear so far.
Dear Jan, Do you have a tutorial regarding 3D Descriptors using RDKit?
Dr. Jensen,
which is better between the Tanimoto similarities and the DICE similarities?
for example I will do molecular docking,
First I have to match the test compound (test ligand) to the natural compound (native ligand) of the protein
The 2nd test ligand with the best similarity index will be carried out by docking the protein molecule
I'm glad you will upload a video explanation to compare Tanimoto Vs DICE using rDKit,
Hello Dr. Jensen, may I ask what is the basis of the nBits=1024 and is it possible to expand this function so I can compare the bitvectors of all indices in the tuple with each other (like you compare index zero with the rest then move to index 1 with the rest and so on)? I have over 500 compounds in my dataset that needed to be clustered and the molecular weight is 1kDa (I am hoping to have a full picture of compounds with similar binding efficiency to a receptor pocket. Thank you very much.
If I understand your question correctly, you want to calculate the Tanimoto similarity between all pairs of molecules. That's certainly possible, just make a double for loop for all your molecules. If you get stuck, consider using the RDKit mailing list: sourceforge.net/p/rdkit/mailman/rdkit-discuss/ . There's a lot of smart people there who are usually happy to help.
I also subscribed to the rdkit mailing list. I watched your video because the RDKit documentation is loaded with too much info and I would like to commend you for making a very simple and easy to understand explanation of how to perform Tanimoto similarity because I also plan to use the same. I just asked if it is possible to perform this on large data sets that will scan the full fingerprint like what you did to Cysteine and Glycine.
May I know what is the basis for 1024 number of bits?
@@nutmitchiko I am happy to hear you found the video useful. You can use any number of bits. 1024 seems to be very popular but I am not sure why
Hello Dr. Jensen,
May I ask how exactly "GetMorganFingerprint" tell us? From what I only know is that any non-zero value indicates that there is a neighbor around, what does "0" tell us, and how to know which atom is the center one?
The fingerprint code makes fragments of various sizes and centered on every atom in the molecule. A "0" and a "1" indicate the absence and presence of a particular fragment in the molecule
How to find whether a compound is aromatic or not?
You can search for aromatic atoms "[a]"
how to find whether compound has a aromatic ring or not?
i have an excel sheet with list of compounds and i need to iterate through all to find whether they a have aromatic rings or not?
Hello sir
How to convert binary fingerprint into their respective smiles?
It's not strictly possible to convert a fingerprint to a SMILES string. Some people have trained machine learning models to to this, but it's still experimental.
Jan Jensen
Sir i am new to ml I have dataset , in which molecule is hashed in to 1024 bit could you please give me suggestions to build a model to predict weather the molecule is toxic or non toxic
Thank you
I appreciate!!! Thank u!
Very helpful
Thank you!
I love the RDKit series. If you have more - please upload! :) Thank you
Hey I am a student(engineering ) from India and I am currently working a project for helping the school students for conducting their chemistry exam through a platform so i need more information about the methods that can be used to verify the structures drawn by the students. So I need your support can I personally contact u or plz provide me your email address for asking some doubts .