• From Nucleotide to Clinic: Computational Paths for Early Detection of Pancreatic Ductal Adenocarcinoma via Liquid Biopsy
  • Zahra Ghanibeygi,1,*
    1. Department of Biology, Faculty of Biology, Falavarjan Branch, Islamic Azad University, Isfahan, Iran


  • Introduction: Pancreatic Ductal Adenocarcinoma (PDAC) has devastating consequences and is usually diagnosed too late to be treated effectively. In the effort to develop an early warning system, scientists have directed their attention to liquid biopsy, which is a simpler blood test that can detect molecular remnants of tumor DNA. The challenge not to simply detecting the DNA fragments, but differentiating beings and subtle signals of early stage cancer from an enormous flood of normal DNA. The plan of this review is to explain how new, complex computational tools based on machine learning and artificial intelligence will help scientists manage these fundamental biological problems by interpreting DNA information as useable human knowledge in the clinic.
  • Methods: My review has mainly focused on recent advancements in utilizing computational tools for detecting PDAC through blood draws. I have presented three main paradigms) Finding Rare Mutations: How new software programs use molecular barcoding to adjust for misses, and find the true cancer mutations that have maybe missed or, even rarer) Using DNA Fragments: How a machine learning model utilizes the floating fragments of DNA not to find mutations, but to identify through the precise random arrangement of fragments as they were broken apart in the bloodstream) Finding Chemical Switches: How computer software finds differences in the epigenetic elements of the cancer, specifically looking at DNA methylation, which are like chemical switches of genes that turn cancer cells on or off. We focused mainly on the most matured modality and looked for ways to combine all of these signals into one highly accurate diagnostic classifier.
  • Results: These results are promising. Computers trained to identify patterns of fragmentation in DNA, can identify early stage PDAC and give us a possible way of identifying the disease rather than rare mutations. There are even algorithms that can identify cancer specific methylation signatures to provide another means of detection, still with good specificity and accuracy. The most progress has been made using a combination of these approaches. Integrated computational models incorporating fragmentation, mutation and methylation data are currently showing incredible capability in identifying early disease that are unprecedented in any of the methods alone. Multi tool blood tests are entering the stage of large trials to determine the potential for intervention in the general population.
  • Conclusion: The prospect of diagnosing pancreatic cancer early from blood draw is a goal becoming achievable, largely owing to advances in computational biology. By interpreting attenuated genomic signals to generate distinctive outputs with bioinformatics pipelines, we are beginning to transduce from the lab to the doctor's office. While we have much work to do in order to transition the tests to clinical application, these results are clearly a paradigm shifting move toward the future of being able to diagnose PDAC in time to make a meaningful impact.
  • Keywords: Liquid Biopsy, Bioinformatics, Pancreatic Cancer