Table 2.

Different ML methods related to various tasks in Drug discovery

Method	Element/features	Task	Refs.
MACCS	Molecular fingerprints	Locating anti-cancer molecules	Kadurin et al. (2017)
CNN	Molecular graph	Identifying graph convolutional fingerprints	Duvenaud et al. (2015), Coley et al. (2017)
CNN	Subsequent graph- based features HE-stained tissues	To predict the disease response in lymphocytes	Saltz et al. (2018), Corredor et al. (2019)
CNN	2D chemical structure image	Biological Activity/toxicity	Pu et al. (2017)
RNN	Molecular Graph	Generating molecules with predicted biological activity	Olivecrona et al. (2017)
RNN	SMILES	To predict molecular properties	Pu et al. (2017)
RNN	Molecular fingerprints and protein sequence	Compound protein interaction	Wang et al. (2016)
RNN	SMILES	Generating novel molecules	Olivecrona et al. (2017)
RNAi	501 cancer cell lines	To identify molecular markers for predicting cancer dependencies	Tsherniak et al. (2017)
DNN	Atom pair descriptor and donor-acceptor pair descriptor	To predict molecular bioactivity	Ma et al. (2015)
DNN	Molecular descriptors and protein features	Drug target interactions	Wang et al. (2014)
M-DNN	Molecular descriptor	To predict the chemical descriptors	Mayr et al. (2016)
Univariate Cox regression	Gene expression signatures	It identifies the predicted high-risk subgroup of victims	Zhan et al. (2006)
Gradient boosted regression trees	Genome-wide polygenic scores	To identify high-risks of breast cancer, coronary artery type 2 diabetes diseases in patients.	Khera et al. (2018)
VAEs	RNA sequencing dataset	Differentiating hidden cancer subpopulations with effectively	Sabrina et al. (2019)
Auto Encoder	Fingerprints	Virtual screening	Kadurin et al. (2017)