Investigating Differences between Ancient and Modern Bacterial DNA in Dental Calculus via Autoencoders

• Published in: 2019 21st International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC) pp. 141 - 147
• Main Authors: Szuhai, Iulia-Monica, Bocicor, Maria Iuliana, Pop, Emilia-Loredana, Mircea, Ioan-Gabriel, Mircea, Cristina
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2019
• Subjects: Ancient DNA; Autoencoders; Bacteria; Bioinformatics; Biological system modeling; Biology; Calculus; Computational modeling; Degradation; Dentistry; DNA; DNA sequencing; Machine learning; Machine learning algorithms; Modern DNA
• DOI: 10.1109/SYNASC49474.2019.00027

Deoxyribonucleic acid (DNA) encodes the totality of hereditary information of all living organisms. A plethora of studies in recent years have been dedicated to investigating DNA of various organisms. One of the many research directions focuses on studying ancient DNA molecules sampled from archaeological sites, which may reveal essential insight about lost populations, extinct species, past diseases or bacteria that may have caused epidemics. However, DNA collected from such locations is often subject to degradation and contamination, thus making its analysis even more cumbersome. This paper presents our experiments towards identifying differences between bacterial DNA from ancient and modern samples, which is important information needed by biologists who study ancient DNA. We investigate autoencoders' capability of capturing the internal characteristics of the DNA molecules and of encoding pertinent differences between ancient and modern bacterial DNA. We perform our experiments on more than 89000 ancient bacterial DNA sequences collected from an archaeological site and our results show the potential of this method.