Genetic Analysis - Automated Confidence Value Determination

The CEQ^TM series DNA sequencing software has built in tools to assist the user in determining the quality of the sequence data generated -- essential for sequence alignment and assembly. By building these capabilities directly into the software, confidence for every base is generated without having to export to third party software.

The quality of called bases are assessed using two calibrations which essentially estimate the probability of error, and depending upon how they are viewed may represent either the probability of error or can be transformed to represent the probability of correctness.

1. Call Scores: Estimate the probability of error and are indicated by a linear scale from 0.00 to 1.00 (0.99 representing 1 error in 100). An example of the display is illustrated in Figure 1.
2. Quality Values: Provide a transformed estimate of the probability of correctness (1 - probability of error), and are presented in a log scale from, 0-50. Generally speaking, quality values of 30 - 50 are considered to represent high quality sequence.

These error probability factors are assigned by a method that does not require prior knowledge of the true DNA sequence. The error rate is the actual number of errors in a section of DNA divided by the number of bases in that section of DNA.

Errors in base calling are usually the result of misinterpretation of peaks in a region of the peak trace, but not in the peak itself. Examining the characteristics of the peaks in the vicinity of the erroneous peak reveals the indicators of error. The most effective parameters for detecting base calling errors consider a window of the trace that includes several peaks flanking the one whose base-call is being assessed.

The parameters used by the CEQ series base caller include:

1. Peak spacing - ratio of the largest peak-to-peak spacing to the smallest peak-to-peak spacing in a given window of seven peaks.
2. Peak resolution - the number of bases between the current base and the nearest unresolved bases x (-)1
3. Uncalled/called height ratio in seven peaks - ratio of the amplitude of the largest uncalled peak to the smallest uncalled peak in a given window of seven peaks.
4. Uncalled/called height ratio in three peaks - ratio of the amplitude of the largest uncalled peak to the smallest called peak in a given window of three peaks.
5. An event pair score - events are detected elongated peaks and interpolated bases.

Call scores utilize parameters 1 - 3. Values generally below 0.97 are considered poor.

Quality values utilize parameters 1 - 5 and produce finer detail in the local peak environment. Quality values utilize a logarithmic scale from 0 - 50. Use of a log-transformed error probability facilitates working with error rates in the range of most importance (very close to 0). The quality value q assigned to a base-call is defined by the following equation:

q = -10 x log₁₀(p)

where p is the estimated error probability for that base-call. For example:

• A probability of error of 1/100 of being correct is assigned a quality value of 20
• A probability of error of 1/1000 of being correct is assigned a quality value of 30
• A probability of error of 1/10,000 of being correct is assigned a quality value of 40

High quality values correspond to low error probabilities. Called bases with assigned quality values of 30 to 40 are considered to be very high quality sequence.