Institute of Molecular Genetics and Genetic Engineering, P.O. Box 794, 11000 Belgrade, Yugoslavia. Communicated by Paul Doty, August 14, 1991. ABSTRACT.
Proc. Nati. Acad. Sci. USA Vol. 88, pp. 10089-10093, November 1991 Genetics
DNA sequencing by hybridization: 100 bases read by a non-gel-based method ZAKLINA STREZOSKA, TATJANA PAUNESKU, DANICA RADOSAVLJEVIC, IVAN LABAT, RADOJE DRMANAC, AND RADOMIR CRKVENJAKOV* Institute of Molecular Genetics and Genetic Engineering, P.O. Box 794, 11000 Belgrade, Yugoslavia
Communicated by Paul Doty, August 14, 1991
ABSTRACT Determination of the sequences of human and other complex genomes requires much faster and less expensive sequencing processes than the methods in use today. Sequencing by hybridization is potentially such a process. In this paper we present hybridization data sufficient to accurately read a known sequence of 100 base pairs. In independent reactions, octamer and nonamer oligonucleotides derived from the sequence hybridized more strongly to this DNA than to controls. The 93 consecutive overlapping probes were derived from a 100-base-pair segment of test DNA and additional probes were generated by incorporation of a noncomplementary base at one of the ends of 12 of the basic probes. These 12 additional probes also had a full-match target in one of the control DNAs. The test and one of five control DNAs spotted on nylon filters were hybridized with 83 octamers and 22 nonamers under lowtemperature conditions. A stronger signal in DNA containing a full-match target compared to DNA with only mismatched targets was obtained with all 105 probes. In 3 cases (2.9%), the difference of signals was not significant (95% successful in another test using simulated data set [50-kilobase (kb) DNA, 100,000 probes] containing 5% false-positive and 5% false-negative data points (24). The Factors of Discriminative Hybridization. We have analyzed the hybridization patterns obtained (i) to determine the dominant factor that leads to a decrease of discrimination with some of the probes and (ii) to estimate the possibility of predicting the hybridization efficiency of untested probes on the basis of sequence. The variations in discrimination are primarily the result of hybridization efficiency, which allows visualization of the signal over the background only if the efficiency is sufficiently large. Another important factor is the kind of mismatched target present in the control DNA. We analyzed the influence of these two factors in more detail. First, the amount of hybrid obtained in test DNA with a certain probe, Hfm, was estimated and given relative values ranging between 0.3 and 10. More than 79% of the probes belong to the group with an Hfm value greater than 1 (Table 2). In this group the average discrimination value was 12 and the minimal discrimination value was higher than 3. In 5 out of 25 probes with end-mismatch targets in control DNA, the D value was >10. A possible explanation can be the large destabilization effect of end mismatches other than G/T or G/A (25) or errors in DNA control sequences. In the group of probes with a hybridization efficiency Hfm of
