Fig 1.
Flowchart illustrating the scoring process for determining sequence by microarray.
Process begins with the scan results file (.gal) to extract MFI data for the purpose of determining the best template sequence and the downstream consensus. A. Features with MFI that are greater than the threshold are initially evaluated by a series of local BLAST alignments to a database of complete FMDV genomes to assemble a genome ‘scaffold’. B. Once the scaffold is produced, the process is repeated with the scaffold sequence as the reference, and individual nucleotides are scored and selected for determination of the final consensus sequence.
Fig 2.
Calculations for inferring nucleotide identities at each position compared to a template sequence.
2A. Microarray feature nucleotide alignment. This represents an alignment of candidate feature sequences in relation to the reference scaffold with the individual nucleotides at each position; 2B. Mean-fluorescent intensity scoring of alignment. These are the weighted scores calculated by determining the MFI of the feature divided by the length of the probe to determine the individual nucleotide score. Different colors represent each of four individual nucleotides.; 2C. Final scoring and nucleotide selection. This chart shows sum scores for each of the possible nucleotides for each position of the alignment, with the highest scoring nucleotide being selected for the downstream consensus sequence. Consensus sequences are produced for multiple MFI thresholds.
Fig 3.
Resolved sequence is selected by alignment of the candidate sequences for each of the thresholds. Sequence assembly is performed through selection of the individual resolved (non-ambiguous) nucleotides at the greatest MFI threshold.
Table 1.
Comparison of percent coverage and identity of microarray inferred interpretive sequences as compared to the predicted viral sequence record in Genbank.
Sequence output from microarray was analyzed using the megaBLAST algorithm against Genbank. The identity of the highest scoring result was compared to the identity of the known viral FMD agents as confirmation of correct identification. Genbank accession numbers of the BLAST results (BLAST ID) are included for the expected nucleotide match, and the accession which identifies the sample sequence within Genbank (Sample ID) by the algorithm. All alignments were found to have significant E-scores of 0.0 by BLAST analysis.
Fig 4.
Comparison of specific misidentified nucleotides when comparing a chimeric virus to a “native template” (upper alignment) assembled sequence versus a “synthetic template” (lower alignment) assembled sequence to illustrate predicted indels and nucleotide error call improvement using the synthetic scaffold template.
Fig 5.
Synthetic template scaffold build-up compared to native template selected by the algorithm.
This is the comparison of the components of the synthetic template buildup for a chimeric FMDV, compared to the selected template compared to an individual scaffold template selection.
Fig 6.
Effect of variable Mean Fluorescent Intensity (MFI) on the identification of nucleotides in the consensus output.
(A) △ 85% Power (Correct); ▲ 85% Power (Unidentified); ◇ 70% Power (Correct); ◆ 70% Power (Unidentified); □ 35% Power (Correct); ■ 35% Power (Unidentified); ○ 20% Power (Correct); ● 20% Power (Unidentified). Decreasing average MFI results in an overall reduction of correct nucleotides with an inverse increase in ambiguous nucleotides; seen as a shift in the 50% correct:unidentified ratio to a lower threshold. (B) ▲ 85% Power (Incorrect); ◆ 70% Power (Incorrect); ■ 35% Power (Incorrect); ● 20% Power (Incorrect) Effect of variable MFI on the percent of incorrectly identified nucleotides at different threshold cutoffs.