Assignment 08

Problem 1. On this web page:

https://www.uniprot.org/uniprotkb/P59594/entry#sequences
under “Sequence” you can find the amino-acid sequence of spike protein, “Spike
glycoprotein”. We do not know whether that is truly the relevant portion of the virus
responsible for Covid. Just download that sequence, or a better, more relevant sequence if
you know how to find it. Please use DeepPurpose API to determine binding affinity for
four medications that were in the news at the time of Covid. Those are:
1) Remdesivir,
2) Ivermectin,
3) HydroxyChloroquine and
4) Chloroquine.
You will have to locate SMILE formulas for those medications. Ivermectin appears to
have two components. Examine the affinity for both. Please report your results whatever
they are.

Problem 2. Use the same amino acid sequence for the Spike protein you identified in
Problem 1 and the dataset of Antiviral drugs from the Broad Institute to determine
whether some of Broad drugs could be used (repurposed) for binding (treatment) of
illnesses caused by the above Spike protein.

Problem 3. Consider the attached notebook deep_repurpose.ipynb. Calculate

binding affinity and ADMET parameters for Remdesivir, GS-441524 and
HydroxyChloroquine for 5 different pretrained models. Analyze and report on the
range of values for the affinity and ADMET parameters for those 3 dugs.

Problem 4. Consider the attached notebook demoModelstm.ipynb and the associated

paper “Toxicity Detection in Drug Candidates using Simplified Molecular-Input Line-
Entry System” by Nath Mriganka and Goswami Subhasish. Chose two molecules in the
provided dataset and compare the toxicity as predicted by this model with known toxicity
as described by one of reference sources, like Merck Index. Report your finding.
Determine the toxicity of Remdesivir, GS-441524 and HydroxyChloroquine
using this method. Compare with values obtained in Problem 3. Check for toxicity values
for those 3 molecules in Merck Index or some other Internet resource.

Problem 5. Examine the attached notebook Molecular_solubility.ipynb and the

Medium paper “Predicting Drug Solubility with Deep Learning” ,
https://sullyfchen.medium.com/predicting-drug-solubility-with-deep-learning-b5e48ff61206.
Use Remdesivir, GS-441524 and HydroxyChloroquine and predict the
solubility. Compare the results produced by this model with standard references like Merck
Index. Report on your result.

1
Please, describe every step of your work. Please present all intermediate and the final
results in Jupyter notebook(s) and their HTML and/or PDF image(s). Your notebook(s)
should include all the results, images and comments.

If your notebook(s) contain(s) excessively long outputs, please copy a meaningful and
illustrative number of initial and/or final lines and paste those in a markdown (comment)
cell. Then, clear the long output(s). Thank you.