Time resolved fluorescence imaging of slab gels for lifetime base-calling in DNA sequencing applications

A compact time-resolved near-IR fluorescence imager nas constructed to obta in lifetime and intensity images of DNA sequencing slab gels. The scanner c onsisted of a microscope body with f/1.2 relay optics onto which was mounte d a pulsed diode laser (repetition rate 80 MHz, lasing wavelength 680 nm, a verage power 5 mW), filtering optics, and a large photoactive area (diamete r 500 mum) single-photon avalanche diode that was actively quenched to prov ide a large dynamic operating range. The time-resolved data were processed using electronics configured in a conventional time-correlated single-photo n-counting format with all of the counting hardware situated on a PC card r esident on the computer bus. The microscope head produced a timing response of 450 ps (fwhm) in a scanning mode, allowing the measurement of subnanose cond lifetimes. The time-resolved microscope head was placed in an automate d DNA sequencer and translated across a 21-cm-wide gel plate in similar to6 s (scan rate 3.5 cm/s) with an accumulation time per pixel of 10 ms. The s ampling frequency was 0.17 Hz (duty cycle 0.0017), sufficient to prevent si gnal aliasing during the electrophoresis separation. Software (witten in Vi sual Basic) allowed acquisition of both the intensity image and lifetime an alysis of DNA bands migrating through the gel in real time. Using a dual-la beling (IRD700 and Cy5.5 labeling dyes)/two-lane sequencing strategy, we su ccessfully read 670 bases of a control M13mp18 ssDNA template using lifetim e identification. Comparison of the reconstructed sequence with the known s equence of the phage indicated the number of miscalls was only 2, producing an error rate of similar to0.3% (identification accuracy 99.7%). The lifet imes were calculated using maximum likelihood estimators and allowed on-lin e determinations with high precision, even when short integration times wer e used to construct the decay profiles. Comparison of the lifetime base cal ling to a single-dye/four-lane sequencing strategy indicated similar result s in terms of miscalls, but reduced insertion and deletion errors using lif etime identification methods, improving the overall read accuracy.