PyEpiSIM User Manual

1 Using PyEpiSIM

1.1 Null simulation

s (1.1.1,Figure 1). Five of general parameters, including Cases, Controls, SNPs, Upper bound of MAFs and Lower bound of MAFs, should be specified by users freely. The parameter of Models is set to be the default value 0.

Figure 1 Interface for setting general parameters

s (1.1.2, Figure 2). The parameter of Average of Adjacent LD should be specified as the default value 0. Other parameters in Figure 2 need not to be set.

Figure 2 - Interface for setting LD parameters

s (1.1.3, Figure 3). Six output parameters, including FileName of chromosome data, FileName of SNP data, chromosome data (.txt), chromosome data (.mat), SNP data (.txt), and SNP data (.mat) should be specified by users freely. Repeat simulation number is set to be default value 1.

Figure 3 - Interface for setting output parameters

s Clicking simulate button for null simulation.

1.2 Generation of LD patterns and haplotype blocks

s (1.2.1, Figure 1). This step is the same as the one (1.1.1, Figure 1).

s (1.2.2, Figure 4). The parameter of Average of Adjacent LD should be specified freely by users in the range of (0,1) . Other parameters (i.e., Position of the first SNP, Position of the second SNP, MAF of the first SNP, MAF of the second SNP, and the LD level) can be set one by one by users. These settings will be accepted while clicking the Specify button.

Figure 4 Interface for setting LD parameters

s (1.2.3, Figure 3). This step is the same as the one (1.1.3, Figure 3).

s Clicking simulate button for the generation of LD patterns and haplotype block.

1.3 Computation of existing public models

s (1.3.1, Figure 1). Five of general parameters, including Cases, Controls, SNPs, Upper bound of MAFs and Lower bound of MAFs, need not to be set by users. The parameter of Models should be set as another value, not the default value 0. This value is the number of models which will be computed or embedded into the simulation data later.

s (1.3.2, Figure 5). There are five steps for the computation of public models. First, selecting the order of a public model (1-SNP to 4-SNP) in the drop-down box. Second, selecting a public model in the list box below the drop-down box. The parametric description of the selected public model is then displayed in the bottom list box. Third, clicking Add button for the computation of the public model. If the information Add OK is shown in the right window, the computation is successful. If the information Add Error is shown in the right window, the computation is failed. Fifth, the successful computation public model can be obtained while the simulate button is clicked.

Figure 5 - Interface of the computation of public models

1.4 Computation of Low-order eME Model

PyEpiSIM provides two strategies for the computation of low-order eME models given MAF_k and the population prevalence p(D): one is based on the heritability h², and another is based on the marginal effect size l .

The one based on heritability h² , MAF_k and the population prevalence p(D) are shown in Figure 6.

s (1.4.1_1, Figure 1). This step is the same as the one (1.3.1, Figure 1).

s (1.4.1_2, Figure 6). There are five steps for the computation of eME models. First, selecting the order of an eME model (1-SNP to 2-SNP) in the drop-down box. Second, selecting an eME model in the list box below the drop-down box. The parametric description of the selected eME model is then displayed in the bottom list box. Third, setting parameters, including The proportion of cases, Heritability of the Model, MAFs of the Model. Fourth, clicking Add button for the computation of the eME model. If the information Add OK is shown in the right window, the 4 computation is successful. If the information Add Error is shown in the right window, the computation is failed. Fifth, the successful computation eME model can be obtained while the simulate button is clicked.

Figure 6 - Interface of the first strategy for the computation of low-order eME models

The one based on the marginal effect size l , MAF_k and the population prevalence p (D) are shown in Figure 7.

s (1.4.2_1, Figure 1). This step is the same as the one (1.3.1, Figure 1).

s (1.4.2_2, Figure 7). There are five steps for the computation of low-order eME models. First, selecting the order of a low-order eME model (1-SNP to 2-SNP) in the drop-down box. Second, selecting a low-order eME model in the list box below the drop-down box. The parametric description of the selected low-order eME model is then displayed in the bottom list box. Third, setting parameters, including The proportion of cases, Marginal effect size of the first SNP ( AA - Aa ), MAFs of the Model. Fourth, clicking Add button for the computation of the low-order eME model. If the information Add OK is shown in the right window, the computation is successful. If the information Add Error is shown in the right window, the computation is failed. Fifth, the successful computation low-order eME model can be obtained while the simulate button is clicked.

Figure7 - Interface of the second strategy for the computation of low-order eME models

1.5 Computation of Higher-order eME Model

PyEpiSIM uses one constraint of h² and p(D) to maximize the other constraint, allowing users to define their own heritability or morbidity without having to find a combination of heritability and morbidity.

The result of maximizing p(D) by constraining h² is shown in Figure 8.

s (1.5.1_1, Figure 1). This step is the same as the one (1.3.1, Figure 1).

s (1.5.1_2, Figure 8). There are five steps for the computation of higher-order eME models. First, selecting the order of an higher-order eME model (1-SNP to 4-SNP) in the drop-down box. Second, selecting an higher-order eME model in the list box below the drop-down box. The parametric description of the selected higher-order eME model is then displayed in the bottom list box. Third, setting parameters, including The proportion of cases, Marginal effect size of the first SNP ( AA - Aa ), MAFs of the Model. Fourth, clicking Add button for the computation of the higher-order eME model. If the information Add OK is shown in the right window, the computation is successful. If the information Add Error is shown in the right window, the computation is failed. Fifth, the successful computation higher-order eME model can be obtained while the simulate button is clicked.

Figure 8 - Interface of the first strategy for the computation of higher-order eME models

The result of maximizing h2 with constraint p (D) is shown in Figure 9.

s (1.5.2_1, Figure 1). This step is the same as the one (1.3.1, Figure 1).

s (1.5.2_2, Figure 9). There are five steps for the computation of higher-order eME models. First, selecting the order of a higher-order eME model (1-SNP to 4-SNP) in the drop-down box. Second, selecting a higher-order eME model in the list box below the drop-down box. The parametric description of the selected higher-order eME model is then displayed in the bottom list box. Third, setting parameters, including The proportion of cases, Marginal effect size of the first SNP ( AA - Aa ), MAFs of the Model. Fourth, clicking Add button for the computation of the higher-order eME model. If the information Add OK is shown in the right window, the computation is successful. If the information Add Error is shown in the right window, the computation is failed. Fifth, the successful computation higher-order eME model can be obtained while the simulate button is clicked.

Figure 9 - Interface of the second strategy for the computation of higher-order eME models

1.6 Search of Low-order eNME models

s (1.6.1, Figure 1). This step is the same as the one (1.3.1, Figure 1).

s (1.6.2, Figure 10). There are four steps for the computation of low-order eME models. First, setting parameters of the 2-SNP eNME model, including Search veracity, The proportion of cases, Heritability of the Model and MAFs by users freely. Second, clicking Search button for the search of the low-order eNME model. If there are one or more models displaying in the right-top list box, the search is successful. If there is no model displaying in the right-top list box, the search is failed. Third, selecting a low-order eNME model in the right-top list box. The parametric description of the selected low-order eNME model is then displayed in the right-bottom list box. Fourth, clicking Add button to add the low-order eNME model into the simulation data. Fifth, the successful computation higher-order eME model can be obtained while the simulate button is clicked.

Figure 10 - Interface for the search of low-order eNME models

1.7 Computation of Higher-order eNME models

PyEpiSIM uses two computational strategies to calculate higher-order eNME models.

The strategy of not using heritability but only using prevalence, is shown in Figure11.

s (1.7.1_1, Figure 1). This step is the same as the one (1.3.1, Figure 1).

s (1.7.1_2, Figure 11). The calculation of higher-order eME models involves four steps. Firstly, set the parameters of the higher-order eNME model, including Choosing the order, The proportion of cases, Heritability of the Model(Can only be 0), and MAFs of the Model. Secondly, click the Count button to search for higher-order eNME models. If one or more penetrance values are calculated in the list box in the upper right corner, the calculation is successful. If no value is displayed in the list box in the upper right corner, the search fails. Thirdly, click the Add button to add the explicit rate value to the simulation data. Fourth, the successful computation higher-order eME model can be obtained while the simulate button is clicked.

Figure 11 - Interface of the first strategy for the computation of higher-order eNME models

The strategy of calculating both heritability and prevalence is shown in Figure 12.

s (1.7.2_1, Figure 1). This step is the same as the one (1.3.1, Figure 1).

s (1.7.2_2, Figure 12). The calculation of higher-order eME models involves four steps. Firstly, set the parameters of the higher-order eNME model, including Choosing the order, The proportion of cases, Heritability of the Model, and MAFs of the Model. Secondly, click the Count button to search for higher-order eNME models. If one or more penetrance values are calculated in the list box in the upper right corner, the calculation is successful. If no value is displayed in the list box in the upper right corner, the search fails. Thirdly, click the Add button to add the explicit rate value to the simulation data. Fourth, the successful computation higher-order eME model can be obtained while the simulate button is clicked.

Figure 12 - Interface of the second strategy for the computation of higher-order eNME models

1.8 Embedding multiple epistasis models

s (1.8.1, Figure 1). This step is the same as the one (1.3.1, Figure 1).

s (1.8.2, Figure 2). This step is the same as the one (1.1.2, Figure 2).

s (1.8.3, Figure 3). This step is the same as the one (1.1.3, Figure 3).

s (1.8.4). Embedding epistasis models into the simulation data until satisfying the number of the parameter Models. There are four strategies for embedding epistasis models.

(1) This step is the same as the one computation of existing public models: (1.3.2, Figure 5).

(2) This step is the same as the one computation of low-order eME models: (1.4.1_2, Figure 6).

(3) This step is the same as the one computation of low-order eME models: (1.4.2_2, Figure 7).

(4) This step is the same as the one computation of higher-order eME models: (1.5.1_2, Figure 8).

(5) This step is the same as the one computation of higher-order eME models: (1.5.2_2, Figure 9).

(6) This step is the same as the one search of low-order eNME models: (1.6.2, Figure 10).

(7) This step is the same as the one computation of higher-order eNME models: (1.7.1_2, Figure 11).

(8) This step is the same as the one computation of higher-order eNME models: (1.7.2_2, Figure 12).

s Clicking simulate button for embedding multiple epistasis models.