Consensus calling algorithms

In 1999, DNASTAR software developer Carolyn Allex published a doctoral thesis (Allex CF, 1999), in which she compared several algorithmic methods for consensus calling, including the “Majority” method and her own “Trace Evidence” method (Allex CF et al. 1997). The latter method was a novel approach for generating quality scores and consensus calling based on geometry and quality of peaks in the trace data.

Allex’s analysis indicated that Trace Evidence had significantly better consensus calling accuracy than Majority, even when many of the individual bases had been called incorrectly. Trace Evidence was also more likely than Majority to make the correct call when the base of the well-defined (true) peak was hidden below a high-intensity valley. By contrast, Majority methods often incorrectly called the base that was associated with the valley.

DNASTAR implemented the Trace Evidence algorithm into SeqMan Pro and, later, SeqMan Ultra. It remains the preferred consensus calling method today. The Majority method is now recommended only when data consist of text sequences rather than fluorescence trace data.

In addition, the quality scoring algorithm that was developed for use with the Trace Evidence method is now also used in SeqMan Ultra for variant calling and quality-based sequence end trimming.