ALG: automated genotype calling of Luminex assays

Abstract

Single nucleotide polymorphisms (SNPs) are the most commonly used polymorphic markers in genetics studies. Among the different platforms for SNP genotyping, Luminex is one of the less exploited mainly due to the lack of a robust (semi-automated and replicable) freely available genotype calling software. Here we describe a clustering algorithm that provides automated SNP calls for Luminex genotyping assays. We genotyped 3 SNPs in a cohort of 330 childhood leukemia patients, 200 parents of patient and 325 healthy individuals and used the Automated Luminex Genotyping (ALG) algorithm for SNP calling. ALG genotypes were called twice to test for reproducibility and were compared to sequencing data to test for accuracy. Globally, this analysis demonstrates the accuracy (99.6%) of the method, its reproducibility (99.8%) and the low level of no genotyping calls (3.4%). The high efficiency of the method proves that ALG is a suitable alternative to the current commercial software. ALG is semi-automated, and provides numerical measures of confidence for each SNP called, as well as an effective graphical plot. Moreover ALG can be used either through a graphical user interface, requiring no specific informatics knowledge, or through command line with access to the open source code. The ALG software has been implemented in R and is freely available for non-commercial use either at http://alg.sourceforge.net or by request to mathieu.bourgey@umontreal.ca.

Figure 1. Graphical user interface provided in ALG.

The Graphic User Interface (GUI) provided in the Automated Luminex Genotyping software (ALG) allows effective management of the genotype calling process. The main interface of the GUI (a) is dedicated to input and output file determination and to parameter selection. The confirmation interface (b) is used to verify parameter selection and to run the ALG analysis. The final interface (c) informs the user of analysis completion.

Figure 2. Graphical plot of data clustering.

Data representation is given as a plot of sum of luminosity on a log scale as a function of the normalized luminosity for a C/T SNP genotyped in 91 individuals. The X axis represents the normalized intensity ϕ, whereas the Y axis represents the sum of the mean intensities of both probes, on a log scale. Each point represents an individual genotype and points are clustered in three groups based on genotype: CC (blue); CT (black); TT (green); no-calls are shown in red. The brackets represent the confidence interval boundaries for a type 1 error of 0.01. The silhouette score (0.979) and HWE p-value (0.782) are also reported.

Figure 3. Manual versus automated genotype calls.

An example of manual versus automated genotype calls obtained from the same assay is provided for SNP rs2267437 (a and b), rs828907 (c and d) and rs11685387 (e and f). The X axis represents the normalized intensity ϕ, whereas the Y axis value represents the logarithm of the sum, of the mean intensity of both probes. Automated calls (a, c and e) were obtained using the default parameters of ALG: confidence α = 0.05; minimum luminescent threshold  = 300 and default cut-off definition method. These parameters yield cut-off values of 0.8/0.4, 0.6/0.3 and 0.7/0.2 respectively for rs2267437, rs828907 and rs11685387. Manual calls (b, d and f) were obtained with the following user-defined parameters: confidence α = 0.0001; minimum luminescent threshold  = 300; default cut-off definition method. These parameters yield cut-off values of 0.6/0.3 and 0.7/0.2 respectively for rs828907 and rs11685387. Cut-off values for the SNP rs2267437 were set to 0.9/0.4 after visual inspection of the results.

Figure 4. Genotype calls of the multi-allelic SNP rs2069416.

ALG analysis of the multi-allelic SNP rs2069416. SNP rs2069416 has three alleles A, T and G leading to three independent genotype calling procedures: A vs T, A vs G and G vs T. The procedure in which an individual is analyzed depends on its two most luminescent probes. The X axis represents the normalized intensity ϕ, whereas the Y axis value represents the sum of the mean intensity of both probes on the log scale. Genotype calls are obtained using user-specific parameters of ALG: confidence α = 0.05; minimum luminescent threshold  = 200; default cut-off definition method. These parameters give cut-off values of 0.6/0.3 and 0.7/0.4 respectively for A vs G and G vs T procedures. Cut-off values for the A vs T were set to 0.7/0.3 after visual inspection of the results.