Monday, November 16, 2026Every story links to its original source.

Genome News

Genomics news, traced to the source.

Reading

On This Day in Genomics

DNA sequencing filters cut errors 7-fold but discarded 12.5% of alignable bases

Combining quality filters cut error rates 7-fold while discarding 12.5% of DNA bases that could be aligned to a reference sequence in a comparison of Illumina HiSeq and GAIIx data. Quality filters screen sequence data to remove potentially unreliable results. Published in Genome Biol. on 8 November 2011, the study by Minoche AE, Dohm JC, Himmelbauer H evaluated sequences from two plant genomes and one virus. Overall error rates in HiSeq data were low, but wrong calls of individual DNA letters accumulated in some regions. Only 3% of all error positions accounted for 24.7% of all substitution errors, in which one DNA letter was mistaken for another.

Examining DNA's forward and reverse strands separately revealed error rates of up to 18.7%. The authors concluded that whether an apparent difference appeared on a particular strand helped distinguish sequencing mistakes from genetic differences present at low abundance. This supported previous recommendations to use strand patterns when interpreting such differences.

The comparison used reads, individual stretches of DNA sequence reported by an instrument, ranging from 95 to 150 bases long. It examined how errors varied with the surrounding sequence and how filtering affected the results. The researchers also assessed the reliability of quality values, the confidence scores the instrument assigns to individual base calls.

Insertions and deletions, errors that added or omitted DNA letters, occurred at very low average rates. But in homopolymers, stretches containing repeated copies of the same DNA letter, those rates rose to as much as 2%.

The authors also found that read coverage, the number of reads covering a DNA position, increased with GC content, the proportion of DNA letters that were G or C. The relationship depended on the GC content range examined.

GAIIx and HiSeq data showed slightly different error profiles. The authors said quality filtering was essential to reduce misleading signals introduced into later analyses.

Sources

Back to today in history · Browse the archive