Egyptian God of the Dead, OSIRIS offers an apt acronym for the "semiautonomous" software program developed by NCBI's Steve Sherry and his team to verify rapidly the quality of DNA data—"housands of records in seconds, tens of thousands in minutes"—and spotlight those findings (perhaps 10 percent) whose ambiguity warrants and can only be resolved by human judgment

The members of the NIH team who had worked together for nearly four years to assist in the identification of victims of the September 2001 World Trade Center (WTC) disaster were again called into action in the wakes of Hurricanes Katrina and Rita.

It was only a few months after the last official meeting in June 2005 of the WTC Kinship and Data Analysis Panel (KADAP) that the calls started going out to reactivate the group. Their experience and expertise were needed to meet the similar—and also quite different—challenges of identifying the Gulf Coast victims.

"It was thought that we could hit the ground running—and, actually, we did," Joan Bailey-Wilson said, referring to the seasoned NHGRI-NCBI cohort whose base of local victim-identification operations, when they are away from NIH, has moved from New York City to Baton Rouge, La.

From KADAP to HVDIEG

The lessons learned in the former effort are being adapted to the unique circumstances of the latter; an approach to identifying victims of mass fatalities in general is emerging; and, increasingly, advances in the science of forensic DNA identification are serving to advance the science of medical genetics—and vice versa, say the NIH investigators.

Among the original members of the KADAP and now a part of the Hurricane Victim DNA Identification Expert Group (HVDIEG) are NHGRI's Bailey-Wilson, co-chief of the Inherited Disease Research Branch; Les Biesecker, senior investigator, Genetic Disease Research Branch, and Elizabeth Pugh, director of bioinformatics and statistical genetics at the Center for Inherited Disease Research in Baltimore.

In the NLM/NCBI contingent are staff scientists Steve Sherry and Lisa Forman. Aside from his victim identification work, Sherry runs the NCBI single nucleotide polymorphisms database (dbSNP). Forman spends about half her time on DNA forensics and the other half on the genetics of rare disease.

It was Forman who, in her capacity as a forensics expert at the National Institute of Justice (NIJ), had assisted the New York medical examiner's office in organizing the KADAP in 2001; see "World Trade Center Victim Identification Pushes Frontiers of Forensic Science," The NIH Catalyst, September-October 2002.)

Forman developed close personal and professional ties with her KADAP colleagues over their four-year collaboration, and after the HVDIEG got underway—coordinated by Amanda Sozer, a forensics expert who had served on the KADAP and was working on high-throughput forensics for the state of Louisiana—Forman was recruited to NCBI, where she officially started in early February.

The Birth of OSIRIS

It was in July of 2003 that an NCBI-NIJ interagency agreement enabled Sherry to initiate his quest to develop the DNA quality-assurance tool that would become OSIRIS. And it was Forman's teenage son who suggested that this new entity be named Osiris, after the Egyptian god of the dead. The words encompassed by the letters in the name—open source independent review & interpretation system—fell easily into place.

The goal, Sherry says, was to address the inadequacies of existing machinery and software designed for use in a "pristine laboratory environment with robust samples and no contaminants"—hardly the situation on the ground at the WTC.

"A lesson learned from 9/11," Sherry says, "is that DNA profile data from samples in challenging environments yield suboptimal results." 

"In any mass fatality," Forman observes, "there can be misalignment of identifications. There were false identifications at the World Trade Center early on; they were ultimately rectified, but it was very distressing for all concerned." The OSIRIS software, she says, provides programmatic checks and double-checks and "recognizes the intersections where errors can lead to misidentification." 

Forensic assessment of identity and whole-SNP genotyping to decipher patient illness, Sherry notes, have similar accuracy requirements. "The software we're writing for OSIRIS will dovetail victim identification and genetic variation in illness." 

A mass fatality can introduce noisy signals that are not part of a DNA profile. Conversely, if part of a sample is destroyed, the intensity of a true signal can be reduced to a point below the normal background cutoff. "We've created software that can semiautonomously adapt to the local characteristics of any dataset, provide its own data analysis, and differentiate noise from a real signal," Sherry says.

"We can digest thousands of samples and quickly identify the small handful—10 percent—of gray-area cases that must be set aside for human eyes to review." 

Joan Bailey-Wilson "One of the big lessons for me from [my involvement in] the World Trade Center identifications was that in times of stress, it is very difficult for police officers not trained in family history data gathering to elicit accurate information from family members. Genetic counseling professionals are the people who can elicit accurate information about biological vs. social relationships within familes. Clinical geneticists, who often deal with severe diagnoses, are familiar with crisis counseling and appropriate interviewing techniques."	Steve Sherry "I try to anonymize the data before regarding them scientifically. I work with the digital representation of victims through their DNA profiles. In that way, the objectivity of the science is not compromised I am not confronted with the real names that call up the images of the real people, which for me is unbearable. . . .I am a product of the U.S. investment in science. I went to public universities and was trained in the theory of population genetics. What I am doing now—using the lessons of 9/11 to work on OSIRIS to ease the pain and suffering of families—is as close as I can imagine myself to being a clinical scientist. I cannot practice at the bedside, but I can do this, and this is my return on the public investment in my training—and my reward."	Lisa Forman When Katrina occurred, it was natural for the state crime lab people in Louisiana and Mississippi, who were charged with the responsibility of identifying the dead, to contact the people who had done that at the World Trade Center. . . .That World Trade Center group—I have never seen such synergy, not in any other group I was ever involved in. Each person brought his or her perspective to the table; issues were always evolving [over the four years the group worked together]; it was an amazingly creative process. . . . And now, once again, from the first meeting in November, this current group has the same energy and synergy."	Les Biesecker "At the end of the day, this effort is about helping families to come to grips with the loss of their loved ones and facilitate the grief process. Grief is one of the most fundamental of human emotions and it can be complicated when the loss of a loved one has any degree of uncertainty associated with it (i.e., were they really killed?) or the absence of the physical remains of that loved one. It is our job to untangle the grief process by providing families with unambiguous evidence of the loss to set them on the road  to come to grips with the tragedy and their loss."

Sherry has been assisting the coordinators of the HVDIEG program—Sozer and Tammy Pruet Northrup, the manager of the DNA unit of the Louisiana State crime lab—in readying OSIRIS to provide informatics support.

"We're building up a statistical profile and framework for victim identification that I'm comfortable with,"  Sherry says.

In the "developmental mode,"  OSIRIS is posted at a free and open website. In maintaining and updating the site, "Lisa will write the documentation; I'll do the artwork,"  Sherry says.

Victim Identification Obstacles
In New York and the Gulf Coast

Before Sherry's first trip to Baton Rouge, NHGRI's Bailey-Wilson and Pugh had been at work on the scene helping state officials develop protocols for family history data collection and installing the relevant software for that project.

Their early presence there was a consequence of the havoc wreaked by Hurricane Katrina, which had created a very different set of forensic circumstances from those surrounding WTC victim identification.

In New York, victim remains were often severely compromised. with full DNA profiles unavailable and body parts, rather than intact bodies, discovered in the wreckage. Reliable reference material, however, was abundant—victims' cups, hair samples, toothbrushes, and the like provided by local family members with relatively easily verified pedigrees and accessible DNA samples for comparison to the remains.

Many of Katrina's victims were recovered with intact tissues and articulated skeletons—but without the adjunct reference samples to establish their identity. Katrina had washed away homes and personal effects, disconnected the victims from those identifiable belongings that could have provided the needed match, making family reference material that much more crucial. But many families were broken up, dispersed throughout the region and even the country.

Thousands were reported missing and feared dead in the hurricane's aftermath because people had no idea where their relatives were or how to get in touch with them.

A Stream of Genetics Counselors

It became clear to Sozer and Northrup early on, Bailey-Wilson says, that many victims would require DNA identification and that genetics professionals would be the best people to contact family members, construct family trees, and explain the rationales and logistics of DNA sampling to those biological relatives whose samples would be most helpful.

Bailey-Wilson adds that for its part, NHGRI was pleased to support the activity and also recognized the opportunity the project offered for training clinical geneticists and counselors on staff and, especially, the genetics counseling students in the joint Johns Hopkins University–NHGRI training program.

With NHGRI's Barb Biesecker, director of the training program, Bailey-Wilson has organized volunteer expeditions typically for one-week stays at the Baton Rouge headquarters. NHGRI funded the travel and other expenses of the first sets of volunteer teams, after which NIH funds kicked in to continue the flow. At first only NIH was supplying genetics counselors; in February, the travel of non-NIH volunteers began to be supported by FEMA.

Bailey-Wilson trains all the volunteers, regardless of where they come from, typically by conference call. She orients them to the purpose of their work in Baton Rouge, which is to interview family members of missing persons and to ask the questions about family relationships that will distinguish biological from nonbiological members. She explains how forensic specialists use the genotypes of relatives to infer the genotype of the missing person, "like fitting pieces into a jigsaw puzzle." 

The session takes no more than two hours because the volunteers have genetics counseling backgrounds. The time it takes for the volunteers to track down family members and to interview them, however, "varies immensely,"  she says.

With extended families or those widely scattered by the storm who have moved multiple times and may be staying in hotels or trailers, there's a lot of detective work involved.

Once the family structure has been determined and the DNA samples agreed to, the geneticists' work is done, and the state arranges for the samples to be taken wherever the individuals are.

DNA samples are genotyped for a set of highly polymorphic genetic loci traditionally used in forensic analysis; it's the job of the statisticians on the HVDIEG, Bailey-Wilson notes, to establish stringent criteria for evaluating match probabilities.

She's learned a lot, she says, from her experiences on KADAP and HVDIEG—about cutting-edge genotyping methodologies, how to deal with degraded samples, and special statistical methods for dealing with unique problems that can arise in trying to impute the genotype of a missing or deceased person from a pedigree. "This work," she notes, "has a direct impact on my own cancer genetics research  (see "Marriage of the Minds.").

There's also the less academic and more joyous rewards of helping families reunite and of being the bearer of good news to distressed people.

Most of the thousands of people reported missing by family members who called in to the Baton Rouge Find Family Call Center have been found alive. "Just about every geneticist who has worked there, in the process of tracking down family members, has found one or two people reported missing who are alive. Every time someone is found alive, the finder rings a special bell, and everyone in the Center applauds. That little bell rang very frequently in December,"  Bailey-Wilson recalls.

"But the families being interviewed now," she adds, "are those whose missing person still has not been found, and most will need a DNA identification and most, it's my impression, are desperate." 

Tomorrow

One day in February, Sherry accompanied two FEMA officers on their rounds in the devastated Lower 9th Ward of New Orleans. Their task was to search the premises at addresses of people reported missing who had yet to be found. They were looking for remains.

They came upon three kinds of scenes: empty space where once there had been an address, standing homes that had been cleaned up by family members, and standing homes filled to the brim with rubble so dense as to make passage difficult—the kind of scenario that could harbor a missing person.

In fact, in such a home only the day before had been found the intact pajama-clad remains of a man in a bed buried under heaps of debris.

So long as such findings are possible, the issue of razing all the uninhabitable residences is problematic, Sherry remarks. Bulldozing, he says, could crush, disarticulate, and blend hidden remains, complicating the ensuing identification efforts.

It has been projected that HVDIEG objectives will have been reached within the year, and Sherry will continue providing his services as long as they are needed. With regard to OSIRIS and assuring the quality of genotype data, he says, "there is no severance date—that's an open-ended commitment."

A similar commitment has been made by the New York medical examiner's office regarding the disposition of remains tagged but not yet identified—a little less than half of the WTC victims. The remains are dessicated and vacuum sealed for maximum preservation, to be entombed collectively in a memorial  structure—and, says Sherry, ever available for identification should new technology emerge that can extract meaningful DNA fragments from material now unreadable.

READING THE SMALL PRINT TO ANSWER SOME BIG QUESTIONS

Jim Ostell There are many reasons for  NCBI's commitment to aid in the identification of the Gulf Coast hurricane victims, not the least of which is that the work contributes to the center's scientific mission, says Jim Ostell, chief of the NCBI Information and Engineering Branch and the person who okays the Gulf Coast activities of the involved NCBI scientists.  Aside from those motives based in the immediate humanitarian imperative, there are the extraordinary scientific rewards of pursuing the research that builds the sequence-analysis tools used by basic scientists. Beyond that, these tools are moving data from the computer screen into the realm of human health. "It's not just academic any more," observes Ostell. "We're moving into a stage at which the molecular biology  of the sequences of the human genome—and of bacteria and viruses —are becoming tools in medicine." One of the first applications of bioinformatics was in refining the tissue typing data upon which bone marrow registries rely for matching donor and recipient. As sequence databases expand and more allele subtypes emerge, standing registry data become less than optimal. Original data from the sequences used in any tissue typing test kit can be run through the NCBI database and signals of otherwise hidden little bits of DNA can be amplified—or not—for more precise characterization. "Without going back to the human involved, taking more bone marrow, or endangering the recipient, we can instantly increase the likelihood that a transplant will succeed," Ostell notes. In collaboration with NIAID, NCBI is sequencing the genomes of flu viruses isolated from patients and tracing their molecular evolution. "We can see how viruses are related to one another, derived from one another." Through understanding the natural history of viruses of interest and maintaining surveillance, scientists can predict the rate at which a virus will likely evolve into an entity against which a vaccine would be warranted. Viral surveillance, he notes, has implications not only in the traditional public health domain but also in the bioterror field that has become part of the landscape. And specifically related to problem solving in the identification of victims of mass disasters has been the development by NCBI's Steve Sherry of OSIRIS (open source independent review & interpretation system). Conceived in the ashes of the World Trade Center and under development for the past three years, OSIRIS is being tested on the ground in the Gulf Coast and is still evolving, according to Sherry (see "Forensics Meets Medical Genetics . . .,"). OSIRIS, says Ostell, is a prime example of an advanced informatics research tool to analyze signal variations that will exert wide impact, ensuring that sequencing and mapping machines yield high-quality data, with particular use in forensics and medical genotyp-ing. NCBI, Ostell observes, has grown in tandem with the bioinformatics field. Established as a part of NLM in 1988, with 12 people, no website, and CD-ROM as the means to release the first databases generated, it now boasts "two million unique users daily, with peak rates of 2,000 web hits a second." It is at the epicenter of translating the data amassed from the Human Genome Project and has served as a first responder to the DNA identification demands arising from manmade and natural disasters. "In the event of future mass fatalities,  Ostell notes, "there will be some new wrinkles, yes, but the response will be more routine because the protocol has been established." He hastens to add, however, that "there's quite a bit left to do—there's always something else."  —Fran Pollner

Return to Table of Contents