1 of 33

Machine Learning in Leukemia Treatment: QSAR Modeling for Lead Compound Identification

K. Sohith Reddy

BE Bioinformatics

SIMATS , Chennai

Tech Talk

2 of 33

Agenda

    • What is Leukemia? What causes Leukemia?
    • How to target Leukemia causing protein?
    • What is Computational Drug Discovery?
    • How does ML aid in CDD? for Leukemia?
    • Overview of my WebApp and identification of lead compound
    • Evaluation of drug-likeliness of lead compound

3 of 33

What is Leukemia?

    • Otherwise, called “Blood cancer”
    • Results in large production of abnormal or immature leucocytes
    • Acute Myeloid Leukemia (AML) is the most prevalent form
    • Abberant differentiaton & proliferation of HSC
    • Results in overproduction of neoplastic clonal MSC

4 of 33

Impact of Leukemia

    • Globally, incidence rate & mortality rate: Males > Females
    • Incidence Rate: 6.1 per 100,000 for Males
    • Incidence Rate: 4.3 per 100,000 for Females
    • Mortality Rate: 4.2 per 100,000 for Males
    • Mortality Rate: 2.8 per 100,000 for Females

5 of 33

So What Causes Leukemia?

6 of 33

Tyrosine Kinase!

7 of 33

Current strategies to treat leukemia

Drug

TK Targets

Midostaurin

Protein Kinase C, FLT3-ITD, TKD

Gilteritinib

FLT3-ITD, TKD

Crenolanib

FLT3-ITD, TKD

Lestaurtinib

JAK2 WT, FLT3-ITD

Sorafenib

FLT3-ITD, RAF, PDGFR, VEGFR, c-KIT

8 of 33

Could there be other drugs that are effective than this?

9 of 33

Prepping and testing numerous drugs for disease treatment is really expensive, isn’t it?

10 of 33

Computational Drug Discovery

    • ML algorithms and computational tools for compound screening
    • Identifying the optimal compound
    • Lead compound selection
    • Transition to biotech companies for drug development
    • Clinical testing and trials
    • Market release of drugs

11 of 33

12 of 33

13 of 33

14 of 33

15 of 33

16 of 33

Overview of my webapp

17 of 33

Bar plots of bioactivity against Lipinski Descriptors

18 of 33

Bar plots of bioactivity against Lipinski Descriptors

19 of 33

Bar plots of bioactivity against Lipinski Descriptors

20 of 33

Comparison of Machine Learning Algorithms

21 of 33

Comparison of Machine Learning Algorithms

22 of 33

Ok, enough of graphs!! So, what’s the inference?

23 of 33

Random Forest Regressor can be the

best ML algo for CDD!

24 of 33

drugs4tyrosinekinase.streamlit.app

25 of 33

26 of 33

27 of 33

28 of 33

29 of 33

30 of 33

31 of 33

32 of 33

Binding Affinity: -7.9

N-(2-chloro-6-methylphenyl)-2-[(2,6-dimethylpyrimidin-4-yl)amino]-1,3-thiazole-5-carboxamide

33 of 33

Thank you!

Please reach out to me on:

Email: sohith.bme@gmail.com

Instagram: heyitssohith

​