Page 275 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

266 Chapter 9 Applications of deep learning in biomedical engineering

mechanisms. The chief biological activities including immunity,
metabolism, signaling, and gene expression are performed via
protein interactions. The larger number of genes involves
complex interactions with some cellular molecules. This leads
to the structural and functional abnormalities in the genes,
which results in a number of diseases. The protein interaction
networks are used to identify genes and proteins correlated
with the diseases.
With the emerging technology, DL algorithms are imple-
mented for predicting protein interaction in the following
applications:
1. Network pharmacology
2. Drug discovery
3. Drugetarget identification (DTI)
4. Elucidation of protein functions
5. Drug repositioning [36]
The PIP comprises three orientations:
1. Proteineprotein interactions (PPIs)
2. DTIs
3. Compoundeprotein interactions (CPIs) [2]

42.1 Proteineprotein interactions
It is the process of at least two protein molecules that work
together to perform highly specific organic procedures. The
establishment of PPIs in overall molecular structure and complex
enzymes is influential. DL algorithms are used for predicting PPI
and act as an emerging field for drug discovery [37].

42.2 Drugetarget interactions
It describes the association of chemical compounds and the
protein targets in the human organism. The applications of DL
advance the DTI in various aspects such as virtual screening,
target prediction, side effect prediction, and drug repositioning
[19]. The working mechanism of DTI is shown in Fig. 9.12.
CPIs: The association between compounds and proteins plays
a significant role in the following applications:
1. Virtual screening for drug discovery
2. Understanding side effects of existing drugs
3. Chemical structure similarity of drugs
4. Sequence similarity of proteins
5. Semantic similarity of proteins
6. Protein domain similarity [38]