ARIADNA: Machine learning method for ancient DNA variant discovery

Joseph K. Kawash, Sean D. Smith, Spyros Karaiskos, Andrey Grigoriev

Research output: Contribution to journalArticlepeer-review

6 Scopus citations


Ancient DNA (aDNA) studies often rely on standard methods of mutation calling, optimized for high-quality contemporary DNA but not for excessive contamination, time- or environment-related damage of aDNA. In the absence of validated datasets and despite showing extreme sensitivity to aDNA quality, these methods have been used in many published studies, sometimes with additions of arbitrary filters or modifications, designed to overcome aDNA degradation and contamination problems. The general lack of best practices for aDNA mutation calling may lead to inaccurate results. To address these problems, we present ARIADNA (ARtificial Intelligence for Ancient DNA), a novel approach based on machine learning techniques, using specific aDNA characteristics as features to yield improved mutation calls. In our comparisons of variant callers across several ancient genomes, ARIADNA consistently detected higher-quality genome variants with fast runtimes, while reducing the false positive rate compared with other approaches.

Original languageEnglish (US)
Pages (from-to)619-627
Number of pages9
JournalDNA Research
Issue number6
StatePublished - Dec 1 2018

All Science Journal Classification (ASJC) codes

  • Molecular Biology
  • Genetics


  • Ancient DNA
  • Genome variants
  • Machine learning


Dive into the research topics of 'ARIADNA: Machine learning method for ancient DNA variant discovery'. Together they form a unique fingerprint.

Cite this