July 1, 1999-June 30, 2001
Level of Access
An investigation into improving the performance of DNA base calling algorithms was conducted. The results have shown that the preprocessing steps performed by ABI sequencer on raw data adversely affects the accuracy of DNA sequencing. This adverse effect has been responsible for relatively high error rates, between 3.5% to 6%, in both ABI and Phred sequencing software. Please note that Phred also uses the processed data generated by ABI sequencer; only their base-calling algorithm is different. To remedy this effect, we have developed and implemented a new filtering technique that preserves the initial information contained in the raw data. This provides qualitatively superior data for the future base calling step. Our proposed filtering step provides mechanical shift compensation, cross-talk filtering, and baseline adjustment. These have been briefly described below. Application of our filtering step on a limited number of DNA data has provided sequences with lower error rate.
Musavi, Mohamad T., "An Intelligent System for Automated DNA Base Calling" (2000). University of Maine Office of Research and Sponsored Programs: Grant Reports. 262.