Limits...
Targeted resequencing of HIV variants by microarray thermodynamics.

Hadiwikarta WW, Van Dorst B, Hollanders K, Stuyver L, Carlon E, Hooyberghs J - Nucleic Acids Res. (2013)

Bottom Line: Seven coded clinical samples (HIV-1) are analyzed, and the microarray results are in full concordance with Sanger sequencing data.Moreover, the thermodynamic analysis of microarray signals resolves inherent ambiguities in Sanger data of mixed samples and provides additional clinically relevant information.These results show the reliability and added value of DNA microarrays for point-of-care diagnostic purposes.

View Article: PubMed Central - PubMed

Affiliation: Flemish Institute for Technological Research, VITO, Boeretang 200, B-2400 Mol, Belgium, Institute for Theoretical Physics, KULeuven, Celestijnenlaan 200D, B-3001 Leuven, Belgium, Janssen Diagnostics bvba, Turnhoutseweg 30, B-2340 Beerse, Belgium and Theoretical Physics, Hasselt University, Campus Diepenbeek, Agoralaan - Building D, B-3590, Diepenbeek, Belgium.

ABSTRACT
Within a single infected individual, a virus population can have a high genomic variability. In the case of HIV, several mutations can be present even in a small genomic window of 20-30 nucleotides. For diagnostics purposes, it is often needed to resequence genomic subsets where crucial mutations are known to occur. In this article, we address this issue using DNA microarrays and inputs from hybridization thermodynamics. Hybridization signals from multiple probes are analysed, including strong signals from perfectly matching (PM) probes and a large amount of weaker cross-hybridization signals from mismatching (MM) probes. The latter are crucial in the data analysis. Seven coded clinical samples (HIV-1) are analyzed, and the microarray results are in full concordance with Sanger sequencing data. Moreover, the thermodynamic analysis of microarray signals resolves inherent ambiguities in Sanger data of mixed samples and provides additional clinically relevant information. These results show the reliability and added value of DNA microarrays for point-of-care diagnostic purposes.

Show MeSH

Related in: MedlinePlus

Analysis of sample no. 3, which is a mixed sample; in contrast to the analysis of unique samples such as shown in Figure 3, in the current case, the resulting  plots always have branches. Thus accordingly, fraction of mismatches higher than the threshold are always found in each iteration. In this Figure, results from analysing t1 (a), t2 (b) and t3 (c) are shown. Notice that the next hypothesis generated from t3, after implementing the nucleotide changes at position no. 12 and 24, is identical to t2. These sequences are shown in Figure 6. Therefore, the algorithm converges into two-cycle state and concludes that the sample in this current case is a mixed sequences sample.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3794611&req=5

gkt682-F5: Analysis of sample no. 3, which is a mixed sample; in contrast to the analysis of unique samples such as shown in Figure 3, in the current case, the resulting plots always have branches. Thus accordingly, fraction of mismatches higher than the threshold are always found in each iteration. In this Figure, results from analysing t1 (a), t2 (b) and t3 (c) are shown. Notice that the next hypothesis generated from t3, after implementing the nucleotide changes at position no. 12 and 24, is identical to t2. These sequences are shown in Figure 6. Therefore, the algorithm converges into two-cycle state and concludes that the sample in this current case is a mixed sequences sample.

Mentions: The plot of Figure 1b is obtained by calculating using a wrong in silico hypothesis: Table 2 shows the actual sequence in solution used in the experiment to produce Figure 1a and the hypothesis sequence made for the calculation of Figure 1b. They differ by a single nucleotide at base position 13 (in the in silico hypothesis this is a G, while the actual target contains a T). Consider now all probe sequences in the microarray with an A at this position. The actual hybridization is a Watson-Crick AT pairing, while the in silico hypothesis estimates the as these were AG mismatches. This leads to overestimating the . The data points corresponding to the probes with nucleotide A at base position 13, are encircled with a solid line in Figure 1b. Conversely, for the probes with a nucleotide C the are underestimated. The latter are encircled with a dashed line in Figure 1b. Thus, the splitting into four branches is due to the wrong estimates of free energy for each probe in the position where actual target and in silico hypothesis differ. It is important to notice here that the probes in the different branches systematically differ from each other by specific nucleotides at specific locations. This systematic sequence deviation will be used to decide whether the plot is branched or not (see right panes in Figures 3 and 5 of the examples in the next section). The correct hypothesis can readily be constructed by selecting out the top left branch, determining the systematic sequence deviation (nucleotide position and type) in this probe subset, and implementing this nucleotide change in the previous hypothesis. This is precisely how a new in silico hypothesis denoted by is generated in block (c) by the algorithm.Figure 3.


Targeted resequencing of HIV variants by microarray thermodynamics.

Hadiwikarta WW, Van Dorst B, Hollanders K, Stuyver L, Carlon E, Hooyberghs J - Nucleic Acids Res. (2013)

Analysis of sample no. 3, which is a mixed sample; in contrast to the analysis of unique samples such as shown in Figure 3, in the current case, the resulting  plots always have branches. Thus accordingly, fraction of mismatches higher than the threshold are always found in each iteration. In this Figure, results from analysing t1 (a), t2 (b) and t3 (c) are shown. Notice that the next hypothesis generated from t3, after implementing the nucleotide changes at position no. 12 and 24, is identical to t2. These sequences are shown in Figure 6. Therefore, the algorithm converges into two-cycle state and concludes that the sample in this current case is a mixed sequences sample.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3794611&req=5

gkt682-F5: Analysis of sample no. 3, which is a mixed sample; in contrast to the analysis of unique samples such as shown in Figure 3, in the current case, the resulting plots always have branches. Thus accordingly, fraction of mismatches higher than the threshold are always found in each iteration. In this Figure, results from analysing t1 (a), t2 (b) and t3 (c) are shown. Notice that the next hypothesis generated from t3, after implementing the nucleotide changes at position no. 12 and 24, is identical to t2. These sequences are shown in Figure 6. Therefore, the algorithm converges into two-cycle state and concludes that the sample in this current case is a mixed sequences sample.
Mentions: The plot of Figure 1b is obtained by calculating using a wrong in silico hypothesis: Table 2 shows the actual sequence in solution used in the experiment to produce Figure 1a and the hypothesis sequence made for the calculation of Figure 1b. They differ by a single nucleotide at base position 13 (in the in silico hypothesis this is a G, while the actual target contains a T). Consider now all probe sequences in the microarray with an A at this position. The actual hybridization is a Watson-Crick AT pairing, while the in silico hypothesis estimates the as these were AG mismatches. This leads to overestimating the . The data points corresponding to the probes with nucleotide A at base position 13, are encircled with a solid line in Figure 1b. Conversely, for the probes with a nucleotide C the are underestimated. The latter are encircled with a dashed line in Figure 1b. Thus, the splitting into four branches is due to the wrong estimates of free energy for each probe in the position where actual target and in silico hypothesis differ. It is important to notice here that the probes in the different branches systematically differ from each other by specific nucleotides at specific locations. This systematic sequence deviation will be used to decide whether the plot is branched or not (see right panes in Figures 3 and 5 of the examples in the next section). The correct hypothesis can readily be constructed by selecting out the top left branch, determining the systematic sequence deviation (nucleotide position and type) in this probe subset, and implementing this nucleotide change in the previous hypothesis. This is precisely how a new in silico hypothesis denoted by is generated in block (c) by the algorithm.Figure 3.

Bottom Line: Seven coded clinical samples (HIV-1) are analyzed, and the microarray results are in full concordance with Sanger sequencing data.Moreover, the thermodynamic analysis of microarray signals resolves inherent ambiguities in Sanger data of mixed samples and provides additional clinically relevant information.These results show the reliability and added value of DNA microarrays for point-of-care diagnostic purposes.

View Article: PubMed Central - PubMed

Affiliation: Flemish Institute for Technological Research, VITO, Boeretang 200, B-2400 Mol, Belgium, Institute for Theoretical Physics, KULeuven, Celestijnenlaan 200D, B-3001 Leuven, Belgium, Janssen Diagnostics bvba, Turnhoutseweg 30, B-2340 Beerse, Belgium and Theoretical Physics, Hasselt University, Campus Diepenbeek, Agoralaan - Building D, B-3590, Diepenbeek, Belgium.

ABSTRACT
Within a single infected individual, a virus population can have a high genomic variability. In the case of HIV, several mutations can be present even in a small genomic window of 20-30 nucleotides. For diagnostics purposes, it is often needed to resequence genomic subsets where crucial mutations are known to occur. In this article, we address this issue using DNA microarrays and inputs from hybridization thermodynamics. Hybridization signals from multiple probes are analysed, including strong signals from perfectly matching (PM) probes and a large amount of weaker cross-hybridization signals from mismatching (MM) probes. The latter are crucial in the data analysis. Seven coded clinical samples (HIV-1) are analyzed, and the microarray results are in full concordance with Sanger sequencing data. Moreover, the thermodynamic analysis of microarray signals resolves inherent ambiguities in Sanger data of mixed samples and provides additional clinically relevant information. These results show the reliability and added value of DNA microarrays for point-of-care diagnostic purposes.

Show MeSH
Related in: MedlinePlus