VANGUARD: An Opensource Approach to Predict Cancer Generation and Evolution by the Use of Nanoparticle SERS, Probabilistic Analysis, and Collaborative Data Processing

Overview

Cancer is one of the most important problems for public health. This project tries to explore a new way to reduce its danger: VANGUARD is a multidisciplinary platform for the generation of universal preventive diagnostics by combining the SERS technic (to detect single-nucleotide mutations in short strings using Raman spectra) with pattern recognizing, collaborative data extraction mechanisms and the design of a probabilistic calculation structure, all in an opensource approach. The first phase consisted in obtaining basic research data to work with. I decided to study a commonly mutated fragment of the DNA keeper TP53 gene (codons 243 to 251) related to the most dangerous cancers: lung, pancreatic and colorectal. Four synthetic copies of this fragment were combined with a simple gold substrate and subjected to Raman microscopy, obtaining 7 samples (1 wild, and 2 for each cancer): using only half of the references needed, and making comparations by eye to reduce cost, samples were recognised properly 83.3% of tests, and the percentage composition of the DNA strings was approximated with an average error of 2.23%. The second phase was developed at home, and it consisted in approaching a diagnostic platform divided into three specific frameworks: data extraction and processing (LIBRA), probability mapping and clustering (ATLAS), and sample identification (ARGOS). After creating a sample recognition software based on example-learning (Retina), thousands of procedurally samples were analyzed, giving an accuracy from 90 to 100%. With web crawlers and a collaborative work app I designed, I used data about dozens of genes to design a calculation to know the cancer potential risk, implemented in scalable approaches for Arduino, Windows, Linux, and Android.

Competition history

• ISEF 2018 Translational Medical Science · Entry TMED045

Resources

• ISEF abstract page ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.