👉Watch this video next (How to learn data science in 2021) ruclips.net/video/oR670Txwh88/видео.html Support this Channel 👇👇👇 🌟 Buy me a coffee www.buymeacoffee.com/dataprofessor 🌟 Download Kite for FREE www.kite.com/get-kite/? 👉 Subscribe to this RUclips channel ruclips.net/user/dataprofessor 👉 Join the Newsletter of Data Professor newsletter.dataprofessor.org
Great explanation professor. I'm just finishing a master degree in molecular pharmacology and as you I'm also a biologist. I'm glad to have found your channel. Sincerely, your work inspire me to keep trying in data science.
I think this kind of problem may the first example of a Hybrid Machine Learning-Deep Learning model. Obviously we have the molecule which we can probably use self-supervised representation learning with, but we also have these handcrafted features that describe the chemical property. Hard to disentangle if you would only want to use one set of features or the other (raw molecule representation vs. chemical descriptors).
Thanks for the insights! Absolutely agree, there’s a lot of possibility in store for the future, deep learning and transformers are paving an interesting path, the possibilities to reverse engineer and generate potent compounds is mind blowing
Thanks for watching, will definitely create more, in the mean time please check on the Bioinformatics playlist that I made, there are 17 videos and several are step-by-step tutorials. bit.ly/dataprofessor-bioinformatics
That’s a great way to put it. The learning algorithm will learn to identify which specific molecular fragments (aka substructures, functional groups, etc.) are important for the biological activity. And yes, there may be a lot of redundant information as well which can easily be removed by filtering out features with near zero variance.
You can choose any label you want to predict(ps:you can find the dataset in open-source web or etc),the active and inactive means 1/0,not a regression problem,but neural network treat them in same way,the diffeeence is outputs.
Professor I have two questions 1. What is the difference between the Pharmacophore model and QSAR and how they are connected? 2. How can we validate a Pharmacophore model using ROC curve without using a paid software (like Ligand scout or MOE) ?
Good job Prof! I am actually going to dive into the chemical field with ML, can you tell me where is the best place to do the literature review about it? I need to prep with what is state-of-art and golden standard right now.
I primarily use Scopus and PubMed to search for keywords related to the topics of my research interest. Additionally I also go specifically to journals in the field such as Bioinformatics, BMC Bioinformatics, Briefings in Bioinformatics, Journal of Cheminformatics, Journal of Chemical Information and Modeling, Chemical Biology and Drug Design, Molecular Diversity, Journal of Computational Chemistry, Journal of Molecular Modeling, Nature Reviews Drug Discovery, etc.
Hello! I'm currently conducting research in this area and your videos and expertise are what I'm looking for! I just started and I'm looking to really dive deep into this research area. I would really love to get in touch with you!
Sure, it’s hellodataprofessor@gmail.com, I can also be reached on linkedin, details in the description of the video as well as on Medium where I post data science articles
Hi, I've written a blog post on this topic "How to Use Machine Learning for Drug Discovery" towardsdatascience.com/how-to-use-machine-learning-for-drug-discovery-1ccb5fdf81ad
Hi DataProfessor, Do you think this sort of approach will assisting in getting the the breakthrough in finally understanding how some anaesthetics work on the brain?
Insights can be gain on what molecular features give rise to the observed actions of drugs. Particularly, interpretations can be made by analyzing the feature importance from e.g. Random forest's Gini index or from Shap values which helps to see the contribution of molecular features. Holistically, we can think of this as: 1. Drug → 2. Target Protein → 3. Various Biomolecular Interaction (Biochemical pathways) → 4. Phenotype (Observable end outcome ) Thus, a single machine learning model can help to figure out steps 1 and 2. Whereas the events that happens in steps 3 and 4 is quite complex and requires further analysis.
Super Explanation Sir! I need information regarding my PhD work, If you could help me in this. My topic is basically about 'A Topical formulation made from bacterial secondary metabolites which can be used against Candida'. Do you know how I can find database of Secondary metabolites from Bacteria where I can search metabolites showing activity against Candida and My Plan is by keeping existing anticandidal compounds as reference is it possible to use Molecular docking to see percentage similarity and then selecting compounds from any of the known Database of metabolites from bacteria? Any help will be highly appreciated.
In the case of unlabeled datasets you could at most perform some unsupervised stuff. Maybe some clustering to try to look into patterns, but without experimental data for your structures, there is no predictive capacity possible.
Hi, how can I contact you please? I want to learn python and machine learning for drug design but I don’t know where to start from. I would like to incorporate it in my phd project
Excellent videos by the Data Professor. Feel free to read the following blog paper on Medium website “Apply Machine Learning Algorithms for Genomics Data Classification”. This will help you to understand how to apply Machine Learning algorithms for genomic data classification. This blog paper contains the latest ML/AI technologies applied to human genomic data classification today.
Hello sir, my self Suman N, im from pharma background completed D pharma , B pharma , I have done Data Analyst, AI & ML, could u plz help me out like how to find pharma IT company or Health care sector to starts as an internship and job. Im waiting for your positive response.
👉Watch this video next (How to learn data science in 2021) ruclips.net/video/oR670Txwh88/видео.html
Support this Channel 👇👇👇
🌟 Buy me a coffee www.buymeacoffee.com/dataprofessor
🌟 Download Kite for FREE www.kite.com/get-kite/?
👉 Subscribe to this RUclips channel ruclips.net/user/dataprofessor
👉 Join the Newsletter of Data Professor newsletter.dataprofessor.org
Do you have a complete course on "Machine Learning for Drug Discovery"?..Beginners to advance
Thnaks bro for ur suport .
Great explanation professor. I'm just finishing a master degree in molecular pharmacology and as you I'm also a biologist. I'm glad to have found your channel. Sincerely, your work inspire me to keep trying in data science.
Wonderful and welcome to the channel! Thanks for watching! Glad the contents are helpful to you 😊
Very cool data prof! Love the short format!
Thanks Ken! At first, I was going to make a under 60 seconds #Shorts, then it turned out to be over 2 minutes. 😆
Wow I have read almost a dozen of papers... but this short video gave me everything i wanted... Thankyou🎉
Awesome intro to Cheminformatics in drug discovery!
Thank you so much! You are the professor I wish I had! Thank you for making this resources FREE!!!!! T^T I CAN'T THANK YOU ENOUGH!!!
Thank you, professor. I'm interested to study more about this subject.
Thanks Data Professor. A concise and brilliant video. Your channel has been helping in my AI in drug discovery research.
Thanks for the kind words, glad it was helpful 😊
As an upcoming 3rd year BS Pharmacy student, I am really excited to use QSAR for our possible thesis topic, really!
That’s awesome, let me know how it goes 😁
@@DataProfessor I will! Thank you very much! ☺ I'll screenshot this comment, so, I could come back again to it then.
Thank you so much, sir. Excellent explanation!!
Most welcome! 😊
Very simple, thank you. Great summary
Thanks for the kind words!
I think this kind of problem may the first example of a Hybrid Machine Learning-Deep Learning model. Obviously we have the molecule which we can probably use self-supervised representation learning with, but we also have these handcrafted features that describe the chemical property. Hard to disentangle if you would only want to use one set of features or the other (raw molecule representation vs. chemical descriptors).
Thanks for the insights! Absolutely agree, there’s a lot of possibility in store for the future, deep learning and transformers are paving an interesting path, the possibilities to reverse engineer and generate potent compounds is mind blowing
Thank you. I will make a neural network for drug discovery now
Great work Professor, I've been following your computational drug discovery series avidly.
Awesome!
Can please make a end to end project of this. I'm a student, your projects helping a lot to learn. I request. Thank you for great information.
Thanks for watching, will definitely create more, in the mean time please check on the Bioinformatics playlist that I made, there are 17 videos and several are step-by-step tutorials. bit.ly/dataprofessor-bioinformatics
@@DataProfessor Awesome, Thank you ) Have a nice day :)
Thanks this was brilliant
Fantastic explaination!
Is knowing the structure and chemical constituents redundant information?
In a sense the model knows those functionally, just not in an explicit way that humans speak of them.
That’s a great way to put it. The learning algorithm will learn to identify which specific molecular fragments (aka substructures, functional groups, etc.) are important for the biological activity. And yes, there may be a lot of redundant information as well which can easily be removed by filtering out features with near zero variance.
would you mind explaining what you mean bioactivity and what you mean by active or non active? What exactly is the model predicting?/active in what?
You can choose any label you want to predict(ps:you can find the dataset in open-source web or etc),the active and inactive means 1/0,not a regression problem,but neural network treat them in same way,the diffeeence is outputs.
Brilliant, hope to see videos titled like this take off.
Thanks for watching, greatly appreciate it. I also am a fan of your channel, great content over there 😆
@@DataProfessor Thanks so much, that's awesome to hear!
That's so awesome! thanks so much! :D Merry Christmas and Happy New Year! :D
Thanks Sergio, Happy Holidays!
Thank you Professor!
A pleasure, thank you for watching 😊
Professor I have two questions
1. What is the difference between the Pharmacophore model and QSAR and how they are connected?
2. How can we validate a Pharmacophore model using ROC curve without using a paid software (like Ligand scout or MOE) ?
Great content sir. Thanks a lot for sharing!!
My pleasure! Glad you liked it 😊
Good job Prof! I am actually going to dive into the chemical field with ML, can you tell me where is the best place to do the literature review about it? I need to prep with what is state-of-art and golden standard right now.
I primarily use Scopus and PubMed to search for keywords related to the topics of my research interest. Additionally I also go specifically to journals in the field such as Bioinformatics, BMC Bioinformatics, Briefings in Bioinformatics, Journal of Cheminformatics, Journal of Chemical Information and Modeling, Chemical Biology and Drug Design, Molecular Diversity, Journal of Computational Chemistry, Journal of Molecular Modeling, Nature Reviews Drug Discovery, etc.
Hello! I'm currently conducting research in this area and your videos and expertise are what I'm looking for! I just started and I'm looking to really dive deep into this research area. I would really love to get in touch with you!
Awesome, you can find me on LinkedIn and Twitter.
So fond of your work. To much easy to learn. Thanks (Y)
Glad to hear that!
@@DataProfessor Would you like to share your contact?
Sure, it’s hellodataprofessor@gmail.com, I can also be reached on linkedin, details in the description of the video as well as on Medium where I post data science articles
@@DataProfessor Many Thanks!
Great diagram !
Thank you!
Hi teacher. Thanks
You are welcome
@@DataProfessor I have question.
I want learn morganfingerprint and I don't know
Nice explanation 👍
Glad to hear that, thanks for watching!
thanks!!
Hi, thank you very much for this insights, you are fantastic. Do you have a step by step using Qsar in your channel?
Is there any book to learn machine learning for the drug discovery
Hi, I've written a blog post on this topic "How to Use Machine Learning for Drug Discovery"
towardsdatascience.com/how-to-use-machine-learning-for-drug-discovery-1ccb5fdf81ad
Hi DataProfessor, Do you think this sort of approach will assisting in getting the the breakthrough in finally understanding how some anaesthetics work on the brain?
Insights can be gain on what molecular features give rise to the observed actions of drugs. Particularly, interpretations can be made by analyzing the feature importance from e.g. Random forest's Gini index or from Shap values which helps to see the contribution of molecular features. Holistically, we can think of this as:
1. Drug → 2. Target Protein → 3. Various Biomolecular Interaction (Biochemical pathways) → 4. Phenotype (Observable end outcome )
Thus, a single machine learning model can help to figure out steps 1 and 2. Whereas the events that happens in steps 3 and 4 is quite complex and requires further analysis.
@@DataProfessor thanks for such a thorough reply! And for all the content you make freely available
@@stretch8390 You’re welcome, it’s my pleasure :)
Super Explanation Sir!
I need information regarding my PhD work, If you could help me in this. My topic is basically about 'A Topical formulation made from bacterial secondary metabolites which can be used against Candida'.
Do you know how I can find database of Secondary metabolites from Bacteria where I can search metabolites showing activity against Candida and My Plan is by keeping existing anticandidal compounds as reference is it possible to use Molecular docking to see percentage similarity and then selecting compounds from any of the known Database of metabolites from bacteria?
Any help will be highly appreciated.
Hi, I would recommend to look into the ChEMBL database
How to built the machine learning model
How are the datasets built?
How to create the representation?
You mean the illustration? I’ve drawn it on an iPad using the Notability app
@@DataProfessor Thank you!
Imma implement it sooner
Great!
Hi, can you expleain about the proteochemometric model concept too? Thank you for your dedicate work!
Thank you for the explanation. How to deal with unlabeled dataset? Can we use pretrained model? If so give any suggestions.
In the case of unlabeled datasets you could at most perform some unsupervised stuff. Maybe some clustering to try to look into patterns, but without experimental data for your structures, there is no predictive capacity possible.
very nice
Hi, how can I contact you please? I want to learn python and machine learning for drug design but I don’t know where to start from. I would like to incorporate it in my phd project
Excellent videos by the Data Professor. Feel free to read the following blog paper on Medium website “Apply Machine Learning
Algorithms for Genomics Data Classification”. This will help you to understand how to apply Machine Learning algorithms for
genomic data classification. This blog paper contains the latest ML/AI technologies applied to human genomic data classification today.
Bro plz maje a video on computational biology
Hello sir, my self Suman N, im from pharma background completed D pharma , B pharma , I have done Data Analyst, AI & ML, could u plz help me out like how to find pharma IT company or Health care sector to starts as an internship and job.
Im waiting for your positive response.
as usual, it is just the fight between 0 and 1 :D
Haha, thanks for watching!
Wow!
Thanks for watching 😊
Barton Stravenue
Althea River
Johnson Track
Brown Fords
2652 Aliyah Center
Wunsch Landing
Wilson Deborah Wilson Melissa Lopez Mary
Toy Harbors