DNA·VIEW NEWSLETTER #21
March 18, 2008
- Dealing with dropout
Allelic dropout is a common confounding factor when dealing with mixtures or in an (old) body identification by kinship.
I have devised a general approach for dealing with dropout, which currently takes the form of protocols combined with
various DNAVIEW calculations (especially the Kinship dropout analysis). In the future I may enhance
DNAVIEW to incorporate the protocols and/or submit for publication papers explaining the approach. In the meanwhile
I can share the approach with any subscribed users who are interested.
- DNAVIEW Enhancements
- Y-haplotype capabilities
Partly motivated by the Katrina identifications, I've enhanced and streamlined the Y-haplotype capabilities.
- You can import and incorporate your own Y-haplotype population data.
- The # can be used to populate the locus list or (for demo) to choose a random probable allele (as with autosomal profile calculations)
- Y-haplotype calculations can be included in Kinship calculations.
I've found a very nice approach to deriving Y-haplotype matching odds, and plan to submit a paper soon.
The Genemapper Import wizard is nicely improved, and includes importing peak-height information,
although no program uses the information as yet. Paternity/Kinship/Crime Case include a comment
field and a useful comment-searching capability so that you can find and/or open cases based on the arbitrary comment.
The kinship module has allowed simultaneous comparison of three or more scenarios for a while,
and this has many uses when investigating the identification of two relatives (consider either or both),
various combinations of odd-man-out when considering three putative siblings, and father/uncle problem among many.
Many new menu items/features include
- Bayes' calculator
- Export symbolic genotype chart with frequencies and allele letters
- Chart of posterior probabilities and "requisite" prior
- calculate one locus showing debugging details
- Show pairwise relationships including parentage, sibling, half-sibling, and cousin.
Please note the "nice user interaction features" below.
Bad bones Kinship dropout
Dropout [NOT] allowed is a special new feature by which,
for user-specified samples ("bad bones"), homozygous genotypes may be analyzed on the assumption an allele is missing.
The usefulness of this feature comes from making two kinship calculations when dealing with bad bones
one with and one without dropout. The correct likelihood ratio is then somewhere in between. (Details forthcoming on request.)
Graphing Kinship Simulations
The nice way to visualize a distribution of likelihood ratios is with a graph. Enhancement to the
Kinship Simulation tool lets you easily do that, even compare several
distributions. Export to Excel, select the data, click Scatter Plot, Finish, and switch to a logarithmic scale.
The new Calculate family trio option in the Paternity case menu
is a convenience which tests for both paternity and maternity. It performs the usual paternity
calculation and then does a maternity calculation for the mother as well. This feature is useful
when dealing with livestock such as horses, where maternity is less certain than is usual with people.
DNAVIEW, DVI and the world
|Paternity index distributions, mother tested or not. (Prof+Cof, Caucasian)
Familial searching searching against a criminal offender DNA database with the idea to find a relative of
a crime scene profile when a direct match isn't there has received a lot of attention lately
(see Finding Criminals Through DNA of Their Relatives, Science 2 June 2006.
Email me if you would like the Supporting On-line Materials which are the scientific meat of the paper.).
The Screen disaster matches module is exactly what one needs to do the searching. I've had some success finding
criminals this way on pilot projects in a couple of jurisdictions.
PATER reports can be in Spanish, if you wish.
- Hurricane Katrina identifications
DNAVIEW is the worldwide standard for mass identification. In 2006 it was the principal analysis tool
for the Katrina identifications, especially those involving kinship (identification with reference to relatives)
of course. I was among the analysts who helped with the casework in Baton Rouge (largest surviving city near New Orleans),
and also served on the scientific advisory panel convened to help sort out technical issues.
Plenty of new ideas arose, supporting the maxim that every disaster is different.
For example, is the prior probability of identity for an anonymous body based on the number of bodies,
or the likely very inflated number of reported missing? More like the latter, but what if, as surely was the case, some bodies were never reported missing?
- Pro-Búsqueda reuniting families
A related application is a relatively new for me project involving reuniting adopted children with their
biological families from whom they were separated (with varying degrees of violence or coercion) during the
El Salvador civil war of the 1980's. The project is called Pro-Búsqueda.
A remarkable case is that of "Juan." Raised in a military household, Juan learned when he was eight
that he was adopted. As an adult he submitted a DNA sample to Pro-Búqueda. A computer search revealed a
cold hit to a family who was hoping for contact with two lost daughters but not a boy. However,
in a follow-up interview the family explained that the mother, who was killed in a military operation,
did have a baby boy but since she was carrying the boy when she died they assumed he was dead as well.
That explanation, consistent dates, and kinship matching odds of 395,000,000 are sufficient to be confident of the connection.
- Growing family
As a follow-on to the tsunami identification project of 2005, I've enjoyed the coincidence and
gratifying experience of meeting people in several different countries who learned how to use DNAVIEW
although neither directly (as is usual) nor indirectly from me. So there is developing a diaspora of DNAVIEW users!
The nice user interaction features below apply to PATER.
Various worthwhile enhancements
- Nice user interaction features
A memory or history of prompt responses is kept so that you can recall your previous responses instead of typing them again.
Use up/down arrow (or PgUp/PgDown
for the multi-line Kinship scenarios) to scroll among previous answers, or ctrl-UpArrow
for a menu. Note also the use of ctrl-z, as in Windows, to restore the original contents of a window.
Another convenience, especially for entering kinship scenarios, is copy/paste.
- Mixture calculations
are a bit easier thanks to the blindly calculate all loci
and default unknown #'s Hp= Hd=. To help cater to dropout, Hd can be a half-integer.
- Case commentsw
help document Paternity/Kinship/Crime cases, and also can be searched to find a case.
- Test cases
is one of several features designed to enhance testing and validation of DNAVIEW.
- Database printout
is clearer and nicer looking.
- Sample switch check
The Paternity Case program gives you a warning if a sample switch looks likely.
- Worklist sizes
Create a report of all profiles for a worklist.
- Notation options
The Options menu allows you to choose international decimal notation for numbers
(99,8% rather than 99.8%), and standard PCR notation (allele 9.3) rather than the symbol which sometimes exports as 9ù3.
- Free tool
Several important enhancements have been added to the installers for DNAVIEW and PATER.
I've made available for downloading
a free installation tool.
It will not alter your current version of DNAVIEW or PATER. It just changes for the better
the way the programs are started (compared to pre-summer-2006) and also supplies a few useful
auxiliary capabilities. The included DNAView Report Viewer.xls is quite useful.
- Installation/Technical fixes
No longer are DNAVIEW and PATER "CPU hogs" (see Free Tool).
This is one of several improvements that are thanks to new installation methods.
- LIMS integration
DNAVIEW has the ability to be harnessed as a slave process to operate invisibly under the
control of your Laboratory Information Management System. There is some custom consulting
involved in establishing the integration. Inquire for details.
- DNAVIEW Latin American seminar
- Early May schedule undecided. Guatemala. All welcome.
- European summer meetings
- I'll be at the following meetings, and in Europe in between
- May 27-30, Ancona, Italy
- June 13-14, Köln, Germany
contact; Home page of Forensic Mathematics