The top 20 chemical compounds leading to cures for age-related disease are identified
Old age is the greatest risk factor for many diseases, so the obvious approach to combat disease would be to slow the aging process. But how? The study of diseases one at a time has proven ineffective, as the disease has already taken hold. It seems that a whole body system approach would be more important in understanding age-related decline. This was the imperative of a group of scientists whose work was described in Aging Volume 9, Issue 7 article entitled, “Machine learning for predicting lifespan-extending chemical compounds” in the search for cures for age-related disease.
Start with selecting the right compounds to study
The study relied on an analysis of data from the DrugAge database, which contains a vast listing of chemical compounds and their effect on the lifespan of organisms studied. The scientists set out to create a way to predict the likelihood that a given chemical compound would increase the lifespan of an organism called Caenorhabditis elegans (C. elegans) with an 80% prediction accuracy rate. By understanding the interaction of a given chemical compound with a living organism, they hoped to predict those compounds in the database which are more likely to increase C. elegans’ lifespan, where the effect of the compounds on an organism’s lifespan is as yet unknown.
The scientists pulled data on chemical compounds from the DrugAge database by searching for the top 20 most important gene characteristics (gene ontology) terms related to cell processes including the parts of the cell where biochemical processes of respiration and energy production occur (mitochondrial processes), and processes that produce certain enzymes related to the immune system, and to terms related to the body’s metabolism and how protein and energy are delivered to cells (transport processes). These compounds were broadly divided into four groups: compounds affecting mitochondria, compounds for cancer treatment, anti-inflammatories, and compounds for therapies that release important hormones in the body (gonadotropin-releasing hormone therapies).
Having a better model leads to better results
The development of drugs is the most practical clinical intervention for aging, since putting people on restricted calorie diets is difficult and hard to maintain, and gene editing is generally unethical in humans. There is a lot of interest in trying to create drugs that mimic calorie restriction, as well as research on new drugs that would slow the aging process. Many of the testing methods for these potential interventions are costly and time-consuming, so the creation of a simple model for testing chemical compounds using the organism C. elegans is very appealing. Although C. elegans is physiologically different from humans, it is the most studied model organism in aging research and produces good results. As well, C. elegans by far has the most known aging-related genes of any model in use (838 at the time of this writing).
The scientists used something called a random forest classification method to classify their findings. This method uses a decision tree that categorizes whether a given result is less possible or more possible. The top 20 selected chemical compound features with the highest average degree of importance are shown below.
|Top 20 selected features with the highest median variable importance.|
|Median Variable Importance||Feature||Feature type||Feature Description|
|14.4||a_nN||MD||Number of nitrogen atoms in the molecule|
|12.8||Isomerase activity||GO||Catalysis of the geometric or structural changes within one molecule|
|11.8||macromitophagy||GO||Degradation of a mitochondrion by macroautophagy|
|11.6||macroautophagy||GO||Process in which cellular contents are degraded by lysosomes|
|11.1||protein disulfide isomerase activity||GO||Catalysis of the rearrangement of both intrachain and interchain disulfide bonds in proteins.|
|11||dipeptidase activity||GO||Catalysis of the hydrolysis of a dipeptide.|
|9.72||pyruvate metabolic process||GO||The chemical reactions and pathways involving pyruvate|
|9.47||PEOE_VSA+4||MD||Total positive van der waals surface area of atoms with atomic charge in the range of 0.20-0.25.|
|9.31||fatty acid transport||GO||The directed movement of fatty acids into, out of or within a cell, or between cells|
|8.79||mitochondrial electron transport, NADH to ubiquinone||GO||The transfer of electrons from NADH to ubiquinone mediated by the multisubunit enzyme known as complex I|
|8.64||vsurf_Wp2||MD||Polar volume at -0.5, a descriptor reflecting the polarizability of a molecule|
|8.57||isotype switching||GO||The switching of activated B cells from IgM biosynthesis to biosynthesis of other isotypes|
|8.4||translation||GO||The cellular metabolic process in which a protein is formed|
|8.18||Q_RPC-||MD||Relative negative partial charge, defined as the most negative atomic charge divided by the sum of all negative atomic charges in the molecule.|
|8.09||aerobic respiration||GO||The enzymatic release of energy from inorganic and organic compounds|
|7.98||a_IC||MD||Atom information content (total), defined as the entropy of the element distribution in the molecule multiplied by the number of atoms.|
|7.95||PEOE_VSA_FPPOS||MD||Fractional polar positive vdw surface area|
|7.86||triglyceride mobilization||GO||The release of triglycerides from storage within cells or tissues, making them available for metabolism.|
|7.79||chi1v||MD||Valence corrected molecular connectivity index (order 1)|
|7.7||bpol||MD||Sum of the absolute value of the difference between atomic polarizabilities of all bonded atoms in the molecule|
Some of the most important features shown in the table above can be related to important and known biological processes of aging and longevity, such as those related to the controlled digestion of damaged organelles within a cell (autophagy) and the processes involved with the creation of energy to run the cell (mitochondrial processes).
Potential implications for human health as we age
To see which top 20 chemical compounds from the database had the highest lifespan-increase potential based on the initial testing, read the study findings in Aging Volume 9, Issue 7. Ultimately, the researchers built, using machine learning, a model to predict the longevity effects of chemical compounds in C.elegans, using the recently published DrugAge dataset. The list of top-hit compounds and their analysis adds to our knowledge of compounds that are most likely to add longevity, and should be researched more in the future. Interventions that slow down the aging process and promote “healthy ageing” could in potentially delay the onset of all age-related diseases, which would benefit human health overall and greatly reduce healthcare costs.