Batch effects in a multiyear sequencing study: False biological trends due to changes in read lengths.

Area of Science:

• Genomics
• Population Genetics
• Bioinformatics

Background:

• High-throughput sequencing (HTS) is vital for biological research but susceptible to batch effects.
• Batch effects, arising from procedural changes, can introduce technical errors and spurious biological signals if not addressed.
• In long-term studies with incremental data, full randomization is often impossible, making batch effects a persistent challenge.

Purpose of the Study:

• To investigate how batch effects can lead to false biological conclusions in multiyear HTS studies.
• To present a case study of spurious selection signals caused by batch effects in Alpine ibex (Capra ibex).
• To highlight risks and discuss mitigation strategies for batch effects in longitudinal HTS data.

Main Methods:

• Analyzed multiyear HTS data from Alpine ibex populations.
• Identified batch effects related to changes in sequencing read length over time.
• Assessed the impact of nonrandom population distribution across different read lengths.

Main Results:

• A change in sequencing read length created a batch effect, leading to false variant alleles and single nucleotide polymorphisms (SNPs).
• Nonrandom distribution of populations across read lengths resulted in allele frequency differences at these false SNPs.
• Biologically spurious signals of selection and environmental associations were detected due to the batch effect.

Conclusions:

• Batch effects pose a significant risk of generating false biological conclusions in multiyear HTS studies.
• Changes in technical parameters like read length can exacerbate batch effects, especially with incremental data.
• Careful consideration and strategic mitigation are crucial to ensure the reliability of findings from longitudinal HTS data.