Iterations

The exact number of iterations that MaSTR runs and records during the MCMC process. As you increase the number of iterations, more of the parameter is explored; however, processing time is increased as a result.

Burn In

The exact number of non-recorded iterations that MaSTR runs before the recorded iterations. You typically set the Burn In value to 20% of the recorded Iterations value.

Thin N

Specifies the in-series step that is to be ignored, for example, every 3rd or 4th step for the defined iteration period. Specifying a thinning value reduces the total amount of data that is analyzed and potentially improves results by reducing auto-correlation. Thinning also improves the “mixing” of the Markov chain (how much movement occurs).

Num Chains

The number of starting points (the initial values for the iterated parameters) for the in-series steps in the MCMC phase of job analysis. For example, if Iterations = 50,000 and Chains = 8, then 400,000 iterations of the in-series steps are being carried out during the MCMC process to calculate the probability values. If the server has the capacity, then you can increase the number of chains, which essentially repeats the iteration process in parallel.

Noise Mean

Used in a model for peak heights that are not accounted for by alleles and stutters. Defined as the mean height of noise peaks in RFUs. If there is a lot of noise, then increase the value.

Presampling Steps

A quick, simplified MCM process is run before the full process to find reasonable starting points. The number that you enter here is the exact number of iterations that are carried for the simplified process. The default value is 10,000 iterations.

You can select one, two, or all three of the following options.

Use Presampling (Allele)

To obtain a better starting point for the allele variables, select this option to run a short MCMC chain on the allele variables by themselves.

Use Presampling (Cont)

To obtain a better starting point for the continuous variables, which can help with convergence, select this option to run a short MCMC chain only on the continuous variables themselves.

Save Iteration Data

Save all the data for each iteration. Leave this option deselected if disc space is a concern or to decrease processing time. Alternatively, in conjunction with using a Random Seed, you should select this option if there is a possibility that you must provide all iteration data for a court case.

Marker 0 LR Replacement

If one of the markers has a likelihood ratio of 0, then specify a value here so that 0 is not the result for the Overall Likelihood ratio.

Theta Value

The value for Theta that is used in the NRCII recommendations.

Population Structure

Select one of three options:

Number of Random Profiles

MaSTR uses the allele frequency data to generate random profiles for each included population. The number of random profiles that MaSTR generates is set here.

Calculate Relatives

Select this option to calculate an additional Likelihood Ratio based on the comparison of the the Person of Interest profile to the profiles of selected relatives.

Use Elimination Database

Select this option to include an Elimination Profile file in the analysis, where an Elimination profile is the profile for laboratory individuals or other individuals whose DNA could potentially contaminate a sample. See Elimination profile.

Elimination Database File

Enabled only if Use Elimination Database is selected. Click Upload to open the File Upload dialog box and select the Elimination Profile file.

LR Flagging Threshold

Any genotype from the Elimination database that has a Likelihood Ratio that falls above this threshold results in a warning being triggered in the Results summary.