Table 1.
Dilution series for one mutation type.
Table 2.
Wild type and mutants sequences.
Table 3.
Wild type and mutants sequences.
Fig 1.
Representative scatter plot: The intensity values from a mixture sample containing a mutant in minority (Imix, y-axis) are plotted versus the intensity of a reference sample containing only wild type DNA (Iwt, x-axis).
A curve is drawn through the reference branch (black circles) using LOWESS smoothing and is called the reference line. The most deviating branch (mutation branch, red squares) contains information about the mismatching nucleotide of the mutant. Probes that do not match the nucleotide of the mutant but have nucleotide variations at the same position can be found in between (side branches). The distance ρ is the difference between the mutation and reference branch and is a visual representation of the change in free energy when a nucleotide mismatch mutation is introduced.
Fig 2.
Scatter plot intensities of the 10C→G KRAS mutant using multiple relative concentrations: The intensity values from the mixed samples (wild type with 10C→G mutant; G12A in amino acid notation) are shown on the y-axis, while the intensity values of the wild type reference are shown on the x-axis.
Data points below background intensity are removed. The plots show a decreasing relative mutant concentration, dilution factor 2.5. The relative concentration, cmut/ctotal, is indicated as a percentage on top of each graph.
Fig 3.
A concentration profile of 10C→G (G12A amino acid mutation) based on Eq 5.
The dashed line is a linear fit with offset zero and a slope which corresponds to exp(−ΔΔG/RT).
Fig 4.
Pie diagrams of the statistical results of experiment set 1: Concentration range experiment with gBlocks samples.
Each diagram corresponds to a relative concentration of the 10C→G mutant (G12A in amino acid notation) in the samples. The size of each pie is proportional to the vertical distance between the mutation branch and reference branch (ρ). The colour of each pie diagram represents the statistical significance: p-values below 0.01 are shown in blue while p-values above 0.01 are shown in orange. All values can be found in S1 Table for reference.
Fig 5.
Plot of concentration profile similar to Fig 3 for all 12 KRAS mutations, based on experiment set 2.
The profile is known to be linear, hence a single experiment per mutation type suffices. A higher slope corresponds to a stronger mutation. The bold dotted line is 10C→G (G12A in amino acid notation), the same mutation we used in Fig 3. The horizontal dashed line corresponds to the threshold value ρt ≈ 0.5 from which we can derive the concentration threshold for each mutation.
Fig 6.
Pie diagrams of the statistical results of experiment set 2: 12 KRAS samples with each mutant present at 5% concentration.
The size of each pie is proportional to the vertical distance between the mutation branch and reference branch (ρ). The colour of each pie diagram represents the statistical significance: p-values below 0.01 are shown in blue while p-values above 0.01 are shown in orange. All values can be found in S2 Table for reference.
Fig 7.
Pie diagrams of the statistical results of experiment set 3: Blinded clinical samples.
The size of each pie is proportional to the vertical distance between the mutation branch and reference branch (ρ). The colour of each pie diagram represents the statistical significance: p-values below 0.01 are shown in blue while p-values above 0.01 are shown in orange. All values can be found in S3 Table for reference.
Fig 8.
A summary of the statistical results of all three experiment sets.
The p′-value of the most significant hypothesis (the true mutation) is indicated by a square. The p′-values of all the hypotheses at the other three nucleotides are indicated with a dot. Raw data can be found in S1, S2 and S3 Tables.