Moreover, it contains the isocitrate binding site and is indispensable for catalytic activity. TARDBP and HNRNPF are both important for gene regulation. TARDBP is normally concentrated in the nucleus but also shuttles between the nucleus and cytoplasm. TARDBP plays an important role in the regulation of splicing, microRNA processing, mRNA transport, stability, and translation. Recent studies showed that TARDBP knockdown inhibited neurite outgrowth and causes cell death. TARDBP dysfunction has been linked to neurological disorders, such as amyotrophic lateral sclerosis, frontotemporal lobar dementia and Alzheimer’s disease. Heterogeneous nuclear ribonucleoprotein F is a member of the HNRNP family that is essential in splicing events. It plays a vital role in modulating gene expression at the transcriptional and posttranscriptional levels. Previous studies have showed that HNRNPF participates at various steps in processing cellular mRNA. In conclusion, we revealed some candidate proteins that might be responsible for the biological differences between CDPSCs and DPSCs. The differently expressed proteins between DPSCs and CDPSCs are mostly involved in the regulation of cell proliferation, differentiation, cell cytoskeleton and motility. In addition, our results suggested that CDPSCs in dental pulp with deep caries have a higher level of expression of antioxidative proteins that may protect CDPSCs from oxidative stress. Further studies are warranted to elucidate the role of potential candidate proteins that may favor dental tissue regeneration. While high-throughput genomic studies have led to the discovery of hundreds and thousands of candidate disease genes, the identification of genes involved in specific human diseases has remained a fundamental challenge, requiring time-consuming and expensive experimentation. Computational approaches that can reliably predict novel disease genes from the vast number of unknown genes will provide a useful alternative to speed up the long and arduous searches for the genetic causes of various human disorders. Given that an increasing number of genes have been experimentally confirmed over the years as causative genes to various human diseases, it will be useful to develop machine learning methods to identify novel disease genes from the confirmed disease genes as positive training examples, based on the observation that genes associated with similar disease phenotypes are likely to share similar biological characteristics. For example, proteins involved in hereditary diseases tend to be long, with more homologs with distant species, but fewer paralogs within human genome. They are also likely to attach together to form functional modules such as protein complexes. In fact, various studies have shown that genes associated with similar disorders tend to demonstrate similar gene expression profiling, high functional similarities and physical interactions between their gene products.