Forensic Mathematics home page
Full user documentation of the mutation feature is in the DNA·VIEW manual
Sample input file for testing mutation computations
Discussion of mutation modelling
Multiple mutations, covert mutations, and false exclusions in paternity casework
STR mutations data from the AABB
STR loci, multiplexes, and DNA·VIEW

STR Mutation Model in DNA·VIEW

Version 25.49 of DNA·VIEW embodies a realistic step-wise model of STR mutation (elaborating on the ideas previously discussed) for paternity (trio) computations – that is, when using Paternity Case calculate report. (The model is not yet implement in Kinship).

Download this sample file and import it, then execute Paternity Case calculate report to demonstrate the mutation computations.

Parameters of the model

The model is controlled by the following parameters:

Invoke the model? Set this Case option parameter "yes" to apply the STR mutation model to STR loci. Set it to "no" to retain the old behavior (the "RFLP model").
In no case will the STR model apply to non-STR loci (such as those coded as SNP or as POLYmarkers).

Rate of paternal 1-step mutations Assuming the STR model is active, the per-locus Probe Parameter "mutation rate" specifies the rate of 1-step mutations from father to child.

Ratio of 1-step to 2-step mutations Two-step mutations are rarer by some factor, perhaps 10 times rarer – or whatever value is specified in this Case option parameter. Three-step mutations are assumed to be rarer still by the same factor, etc.

Ratio of paternal to maternal mutations Set this Case option parameter to 3.5 or so to model the fact that maternal mutations are less common than paternal.

Fraction of mutations that increase size Setting this Case option parameter to 0.5 is good enough. Future implementations may incorporate a more elaborate model to cater to Tim Clayton's claim (for which he cites Xu et al) that the ratio may be significantly unbalanced for large alleles.

Lower bound, probability of any mutation Leave this Case option parameter as 0 normally. It provides a way to cater to an oddball mutation if one is ever suspected.

Minimum PI (other than 0) A value of e.g. 0.01 for this Case option parameter means that if the cumulative effect of mutation calculations drives the overall likelihood ratio below 0.01, treat them all as "exclusions" and report PI=0.

Maximum number of mutation loci in a case A more traditional but less logical approach is to set this parameter to 2 or 3. Recommendation: set this one to 999 and rely on the previous one.

Invoke the model?	Set this Case option parameter "yes" to apply the STR mutation model to STR loci. Set it to "no" to retain the old behavior (the "RFLP model"). In no case will the STR model apply to non-STR loci (such as those coded as SNP or as POLYmarkers).
Rate of paternal 1-step mutations	Assuming the STR model is active, the per-locus Probe Parameter "mutation rate" specifies the rate of 1-step mutations from father to child.
Ratio of 1-step to 2-step mutations	Two-step mutations are rarer by some factor, perhaps 10 times rarer – or whatever value is specified in this Case option parameter. Three-step mutations are assumed to be rarer still by the same factor, etc.
Ratio of paternal to maternal mutations	Set this Case option parameter to 3.5 or so to model the fact that maternal mutations are less common than paternal.
Fraction of mutations that increase size	Setting this Case option parameter to 0.5 is good enough. Future implementations may incorporate a more elaborate model to cater to Tim Clayton's claim (for which he cites Xu et al) that the ratio may be significantly unbalanced for large alleles.
Lower bound, probability of any mutation	Leave this Case option parameter as 0 normally. It provides a way to cater to an oddball mutation if one is ever suspected.
Minimum PI (other than 0)	A value of e.g. 0.01 for this Case option parameter means that if the cumulative effect of mutation calculations drives the overall likelihood ratio below 0.01, treat them all as "exclusions" and report PI=0.
Maximum number of mutation loci in a case	A more traditional but less logical approach is to set this parameter to 2 or 3. Recommendation: set this one to 999 and rely on the previous one.