Powerful Probabilistic Genotyping Tools

The eDNA Consortium provides Enhanced Semi Continuous (Bullet) and Continuous Methodologies (BulletProof) for all your Probabilistic Genotyping and Mixture Deconvolution needs.

Is one method better than the other?  No.  But one or the other may be more suitable for an organization.  The various methodologies are complementary-and facilitate a robust and sensible validation process.  As an eDNA Consortium Member you will have full access to both Bullet and BulletProof.     

BulletProof is the latest addition to the arsenal of Probabilistic Genotyping tools made available to the eDNA Consortium.  

Request full access to BulletProof

Forensic Laboratories in the United States are slow in moving towards Probabilistic Genotyping: Why?

  • The “Commercial” software is too expensive…
    • Join the eDNA Consortium and take advantage of the many free calculation tools
  • The dread of validating a new calculation procedure
    • The eDNA Consortium has developed a comprehensive User Guide, training program, and companion Manual Verification Spreadsheet doing much of the validation work for you.
  • The fear of overly complex calculations—and the ability to adequately testify to the results
    • The eDNA Consortium is cognizant of the “Black Box Fear” when implementing software solutions—Therefore eDNA’s Probabilistic Genotyping (PG) tools include a comprehensive Step by Step calculation manual, numerous Published Papers, and multiple Spreadsheet Solutions demonstrating the PG algorithm in action—and full concordance.
    • Become comfortable using eDNA’s PG Tools with onsite PG training provided by the eDNA Steward.
    • Join the eDNA Probabilistic Genotyping User’s group for sharing information among colleagues.
  • The perception that calculations can take hours or days to perform
    • Unlike Commercial software solutions, Bullet performs calculations in seconds.  The eDNA Probabilistic Genotyping tool has been designated Bullet for its speed, accuracy, and ability to cut through the fog of uncertainty.
  • The perception that results are not reproducible
    • Unlike some MCMC software solutions, Bullet produces identical results each time the same scenario is calculated. When using Bullet the expert will never need to face the question, “Is it true your results are not reproducible?”  

Why is Probabilistic Genotyping necessary.

The trend in the type of samples Forensic Laboratories are asked to resolve, along with improved extraction methods, amplification chemistries, and improved CE instrumentation have exceeded the ability to adequately resolve mixtures using binary mixture interpretation protocols. When laboratories try to analyze highly complex mixtures, such as “touch” items with more than two contributors and stochastic data, binary methods (CPE, CPI, Modified RMP) fail miserably. With traditional binary methods there exists no way to factor uncertainty.

Current strategies to evaluate low-level mixtures with dropout using the binary methods are insufficient. The existing CPE / CPI methods have no fundamental validity, there exists no unambiguous and rigorous mathematical proofs—it is fundamentally flawed and at best, a “stab” at the weight of the evidence.

What is Probabilistic Genotyping?

Years ago, the ISFG published recommendations for the interpretation of low-level mixtures when dropout is possible (Gill et al. 2012). eDNA’s Probabilistic Genotyping tools provides the ability to utilize all data above the Laboratory’s empirically determined analytical threshold by accounting for “uncertainty” wherein we incorporate a probability of allelic drop-out, drop-in (and if warranted, alternate hypotheses including the ability to account for possible “related suspects”) in the LR.

Quoting the 2016 SWGDAM Guidelines for Validation of Probabilistic Genotyping Systems:

“Probabilistic genotyping refers to the use of biological modeling, statistical theory, computer algorithms, and probability distributions to calculate likelihood ratios (LRs) and/or infer genotypes for the DNA typing results of forensic samples (“forensic DNA typing results”). Human interpretation and review is required for the interpretation of forensic DNA typing results in accordance with the FBI Director’s Quality Assurance Standards for Forensic DNA Testing Laboratories . Probabilistic genotyping is a tool to assist the DNA analyst in the interpretation of forensic DNA typing results. Probabilistic genotyping is not intended to replace the human evaluation of the forensic DNA typing results or the human review of the output prior to reporting.”


In the example below the system would be unresolvable using CPI/CPE or other binary LR methods. However Bullet incorporates uncertainty and allows the analyst to use all the data and weigh the evidence accordingly.

Drop-out (and Masking) example on a Two Person Mixture

Bullet allows a User to optionally investigate the Evidence Profile “system by system” to further visualize or investigate. Bullet incorporates each allele’s RFU values with associated Base Pair Size in an explicit visual manner allowing the analyst to calculate/verify “on the fly”  the system’s probability of drop-out. 

Optional Contributor Isolation Workspace

Ultimately, by using all data and automated assignment of an appropriate level of uncertainty (based on Signal Strength / Base Pair sizes) in conjunction with a Degradation Curve, all data from the Crime Scene Profile can be used. The resulting Likelihood Ratio is calculated on competing hypotheses used to explain the presence or coincidental match of a suspected contributor’s profile within a Crime Scene profile.

Example of a two-person mixture conditioned on the stochastic minor contributor

Bullet is further enhanced to easily accommodate mixed samples exhibiting differential degradation. 

Bullet will derive and display the Degradation Curve allowing the analyst to gain valuable insight into the Crime Scene Profile.  The optional curve can then be used to predict the allele’s anticipated Signal Strength based on Amplicon length, further normalizing and allowing the analyst to condition the level of uncertainty in systems where the contributor’s alleles are masked or may have dropped out altogether.    

Degradation Curve Visualization


Once a user becomes comfortable with Bullet (Enhanced Semi Continuous Probabilistic Genotyping model) theory and implementation, the Continuous methodology in BulletProof will not be such a scary leap from CPI and other traditional methods.  This incremental approach facilitates training, deep level understanding, and confidence. 

BulletProof provides a high level of automation. The Maximum Likelihood approach is based on the frequentist inference maximizing the likelihood function with respect to the unknown parameters to obtain the maximum likelihood estimate.

The software optimizes the Likelihood (under each hypothesis) as a function of the unknown parameters in the continuous model:

1)         Mix-Proportion: (mx1,…, mxC): mixture proportion for contributor 1,..,C.

2)        PH Expectation:  mean of a heterozygote peak height allele

3)        PH Variability: coefficient of variance for a heterozygote peak height allele

4)        Degradation: degradation slope

5)         Stutter Prop: (n-1)-stutter proportion

Lab Agnostic

This approach chooses the parameters which provide the best fit to the observed peak heights. Hence the uncertainty of the parameters are not required which makes this a great tool to implement in lab agnostic analysis. The Maximum Likelihood method does not require lab specific “training sets of data”.

This also makes intra-lab data simpler to manage when running multiple chemistries on multiple pieces of equipment. E.g. 3130 vs. 3500.

BulletProof additionally provides the MCMC Bayesian LR, and options to run the Numerical Integration Bayesian methods.

The BulletProof Run Wizard makes it simple to Setup and Queue a Probabilistic Genotyping calculation series.  

Simple Setup

The Organization’s default parameters are presented, allowing users an option to change based on evidence characteristics namely whether to use Degradation and /or Stutter.

The last step is to review the Run scenario and Queue. Users can Queue multiple scenarios and they will be notified when results are available for viewing.  This scenario required 20 Seconds calculation time but complex Runs can take minutes to an extreme of hours.  The results Overview provides the Weight of Evidence and graphics designed to aid visualization of the scenario calculated.      


The user can delve deeper by examining the Hp and Hd hypothesis to include Model Validation, Mixture Proportions, and deconvolution results. 


Request full access to begin using these tools today.

The recent BulletProof AIC release was made available for production use.  Akaike Information Criterion (AIC) is an automated tool incorporated in BulletProof which provides a mathematically rigorous method to score various models to assist the user in determining the best fit model while considering number of contributors, stutter, and degradation. 

Akaike Information Criterion

In this scenario a user would select the Id hyperlink of the top line (the largest AIC Score) to view the result representing the “best fit model”.  

Bullet & BulletProof eDNA’s Probabilistic Genotyping Tools

BulletProof, eDNA’s Continuous Probabilistic Genotyping tool provides a Frequentist and Bayesian methodology to resolve complex mixtures and provide mixture deconvolution.  Click below to schedule a free demo and access to the fully operation application.

Take the First Step

Contact the eDNA Steward to arrange for an online primer demo and for login access to the fully operational demo site featuring eDNA 3.2 with eSolve Indexing System, Brutus, and Bullet.

There will be no sales pitches—just a pure tour of the high-level functionality of the eDNA LIMS…then the user will be turned loose to explore the deeper domain functionality at their pleasure.

The demo login credentials provided during the tour remain valid indefinitely so come back whenever and as often as you wish.