Journal Entry Week 14 20221129 - klmartinez/DSF GitHub Wiki
For this week's Tuesday lecture we are covering the topic of Deep Learning: Overview of Deep Learning Algorithms.
Commonly used python-based general deep learning libraries: TensorFlow & Pytorch
Weekly Challenge
Instructions:
DSF Weekly Challenge (11/30/2022) -Application of Neural Networks in Health Sciences.\
Based on our notes of an Overview of Deep Learning Algorithms, you are asked to search for at least 3 research papers in your field of research that use some deep learning algorithm or list of them. Try using the keywords “deep learning” and your “research keys” in Pubmed, Google Scholar, or other of your choice.\
Then write a short summary of this Challenge, include the citations to the 3-5 papers found, and the specific deep learning neural networks that they are using and what is the research question these algorithms are helping to respond. Question: How would deep learning neural networks would benefit your current research?
Example: Zou, J., Huss, M., Abid, A., Mohammadi, P., Torkamani, A., & Telenti, A. (2019). A primer on deep learning in genomics. Nature genetics, 51(1), 12-18. (URL link) Algorithms used: Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), … Application: Predicting the sequence specificity of DNA- and RNA-binding proteins and of enhancer and cis-regulatory regions, methylation status, gene expression and control of splicing (Made up….)
Summary
My training was in genetics, and my postdoc is specifically focused on the topic of pharmacogenomics so I focused my searches in these two categories. I have found that deep learning has already been used in pharmacogenomics including 1) the identification of novel regulatory variants located in noncoding domains of the genome and their function as applied to pharmacoepigenomics, 2) patient stratification from medical records, and 3) the mechanistic prediction of drug response, targets and their interactions. Deep learning is also being utilized in the broader topics such as in prognosis and survival prediction (ex: for cancer) using multi-omics data including genomic data. DeepProg has been developed to do this and is described as a “novel ensemble framework of deep-learning and machine-learning approaches that robustly predicts patient survival subtypes using multi-omics data”. Deep learning has been applied to genetic variant calling, metagenomics, single-cell transcriptomics, and epigenetics. For example, DeepCpG has been used to predict CpG methylation from DNA sequences using convolutional neural networks. It is anticipated that deep learning will be widely used to predict personalized drug response and optimize medication selection and dosing, using knowledge extracted from large and complex molecular, epidemiological, clinical and demographic datasets. I would very much like to do this as well in our lab as well focused on the cardiovascular drugs that interest us.
Publications:
- Kalinin AA, Higgins GA, Reamaroon N, Soroushmehr S, Allyn-Feuer A, Dinov ID, Najarian K, Athey BD. Deep learning in pharmacogenomics: from gene regulation to patient stratification. Pharmacogenomics. 2018 May;19(7):629-650. doi: 10.2217/pgs-2018-0008. Epub 2018 Apr 26. PMID: 29697304; PMCID: PMC6022084. link
- Poirion OB, Jing Z, Chaudhary K, Huang S, Garmire LX. DeepProg: an ensemble of deep-learning and machine-learning models for prognosis prediction using multi-omics data. Genome Med. 2021 Jul 14;13(1):112. doi: 10.1186/s13073-021-00930-x. PMID: 34261540; PMCID: PMC8281595. link
- Zou J, Huss M, Abid A, Mohammadi P, Torkamani A, Telenti A. A primer on deep learning in genomics. Nat Genet. 2019 Jan;51(1):12-18. doi: 10.1038/s41588-018-0295-5. Epub 2018 Nov 26. PMID: 30478442. link
- Schmidt B, Hildebrandt A. Deep learning in next-generation sequencing. Drug Discov Today. 2021 Jan;26(1):173-180. doi: 10.1016/j.drudis.2020.10.002. Epub 2020 Oct 12. PMID: 33059075; PMCID: PMC7550123. link