Omics based aging clocks for chemical screening

Aging is a progressive and generalized deterioration of the functional capacities of an organism and the leading cause for most chronic diseases (Campisi, 2013). And this aging process results in multiple traceable footprints, for example, changes in gene expression and chemistry concentrations, epigenetic changes, telomere attrition (Carlos et al., 2013; Hernandez-Segura et al., 2018). These biomarkers can be quantified and used to estimate the age of an organism called aging clock (Bocklandt et al., 2011; Horvath, 2013). Today there are dozens of aging clocks based on such biomarkers of aging as DNA methylation, gene expression and metabolomic profiles (Zhavoronkov et al., 2019; Lu et al., 2019; Fahy et al., 2019; Schultz et al., 2020).
Aging clock can dissociate biological from chronological age, and thus it has a wide application in aging research (Zhavoronkov et al., 2019). For example, measuring biological age is important to determine the influence of environment (Horvath et al., 2016), diet or therapies on the aging process (Kresovich et al., 2021), and the development of age-related diseases (Quach et al., 2017). The aging clock also allows researchers to accurately predict the anti-aging effects of gene variants and various chemical treatment (Lu et al., 2018).

Transcriptomic data are one of the most abundant types of data, there are several public databases for RNA sequences, for example, GEO (https://www.ncbi.nlm.nih.gov/geo/), ENCODE (https://www.encodeproject.org), EBI (https://www.ebi.ac.uk), GTEx (https://gtexportal.org/home/) and so on. Transcriptome-based models for drug screening is emerging as a powerful tool to identify compounds to intervene in many different diseases (Vamathevan et al., 2019). This kind of predictors may can be trained to drug-induced transcriptomes, and then used to predict anti-aging compounds that produce “young” transcriptomes. 
The overall purpose of this project is to generate a set of transcriptomes-based age clocks, and use them to screen compounds for healthy aging. If possible, we additionally wanted to do some wet experiments to verify the effects of candidates compounds in vitro. To do this, we will employ machine learning, differential expression, and enrichment analyses in LifeLine-Deep RNA-Seq datasets to train the models and then use them to predict compounds with the LINCS L1000 small molecule expression profiles.