Resolving paternity relationships using X-chromosome STRs and Bayesian networks.

Area of Science:

• Forensic Genetics
• Computational Biology
• Statistical Genetics

Background:

• X-chromosomal short tandem repeats (X-STRs) are valuable genetic markers for paternity testing due to their unique inheritance patterns.
• Autosomal STRs (as-STRs) are commonly used, but X-STRs provide complementary data, especially in complex cases.
• Likelihood ratios (LRs) are crucial for quantifying the probability of paternity, often calculated using complex formulae based on pedigree structures.

Observation:

• Bayesian networks (BNs) provide a graphical and computational framework for calculating complex probabilities in genetic analysis.
• Two specific BNs were developed and validated for calculating LRs in half-sister/half-sister and mother/daughter/paternal grandmother relationships.
• The BNs demonstrated utility beyond their initial design, proving adaptable to other suspect pedigree situations.

Findings:

• The mother/daughter/paternal grandmother BN application confirmed the added value of X-STRs alongside as-STRs in paternity cases.
• Evaluating a case without maternal DNA demonstrated that missing information can lead to conservative or non-conservative LR estimates depending on the alleged father's status.
• The half-sister case highlighted limitations in statistical interpretation when dealing with high allelic frequencies of STR markers.

Implications:

• BNs offer a more accessible and potentially more accurate method for calculating paternity probabilities in forensic genetics.
• The study underscores the importance of utilizing X-STRs in conjunction with as-STRs for robust paternity testing outcomes.
• Further research into BNs could refine statistical interpretations in forensic genetics, particularly concerning allelic frequency variations.