On April 29, a conference call from New York City to Ann Arbor-based Gene Codes Corp. delivered good news of a sort: The software it had built and perfected over the past seven months had identified the 1,000th victim in the Sept. 11 World Trade Center attacks.
Staff members gathered in the programming "pit" to accept congratulations from founder Howard Cash, who was stationed at the Offices of the Chief Medical Examiner of New York City.
"It's sobering but also satisfying that we are contributing to identifying victims in a disaster of this magnitude," Cash said.
The pattern-matching software that the 14-year-old bio-informatics company designed is slowly giving thousands of families closure by trying to match 20,000 remains stored in refrigerated trucks with the 2,780 missing Trade Center victims.
Gene Codes became involved in the investigation Sept. 26 when a former client employed by the medical examiner in New York telephoned Cash and asked whether he could come there the next day.
Cash flew out with a suitcase full of software, prepared to donate it to the identification effort.
Instead, Robert Sheller, director of the Department of Forensic Biology for the medical examiner, asked Cash whether he could build software to manage inventory and identify victims' remains, a process that was proving to be impossible using current technology.
Cash told Sheller that neither Gene Codes nor anyone else commanded the resources to build such a complex system in the necessary time frame: immediately.
Shaler told Cash to hire whom he needed.
Cash returned to Ann Arbor, gathered his 15-member staff and explained the project.
"We're going to have to bring on people very fast, we're going to have to work extremely long hours, we will lose financially on it," he recalled telling his employees.
He stressed the gravity of the undertaking.
"If we misidentify one person, there is one family that will never get over that," Cash said.
He asked whether anyone objected. No one did.
"I think that a lot of people felt that it suddenly became your personal war," recalled Amy Sutton, quality-assurance specialist for Gene Codes.
In the following months, Gene Codes used about $2 million in retained earnings to hire 11 people and more than double office space from 3,400 to 7,800 square feet.
It created a subsidiary called Gene Codes Forensics for the Trade Center project, largely to protect the rest of the company against potential lawsuits that could arise from misidentification or if every victim isn't identified.
Now, the company's entire technical staff, except for two employees, is working solely on the Trade Center project. Development of new products and upgrades has all but stopped.
"The salespeople, who are not working on this project, know that their work is kind of financing the work for the World Trade Center," Cash said.
Nine computers in the Forensic Biology Department in New York run the software, called M-FISys - an acronym for mass fatality identification system, Shaler said. Each Friday, Gene Codes sends a new version of M-FISys to New York, but not before hundreds of quality-assurance tests ensure that the program works. So far, there hasn't been a software-based mistake.
Gene Codes is not getting rich from the venture, although it could.
"If we said we'd do it for $10 million, they'd pay it," Cash said.
But the project is about more than money. Gene Codes has a three-year, $10 million contract with New York City, but Cash said he expects to charge only $2.5 million to $3 million.
"It's a call to service," he said.
Cash feels fortunate that as a privately held company with 12 shareholders, Gene Codes was free to choose the Trade Center project for its public-service value.
Revenue totals about $5 million annually, and Gene Codes has been profitable for the past 35 quarters. But the company expects to lose money this year largely because of expenses associated with the Trade Center project that haven't been reimbursed yet.
Gene Codes plans to recoup those losses through growth in other parts of the business in 2004-2005, Cash said.
He's collected a stack of cards, mainly from other countries, asking for information about the software. He recently received an e-mail from the head of the DNA Laboratory at the National Center of Forensic Medicine in Tel Aviv, Israel, where terrorist attacks have created a need for the identification software.
The emotion surrounding such an undertaking is tough for outsiders to understand.
"You have to find a way to cope. Just the data you're dealing with, it hits you that these are people, and people are waiting desperately to get answers," Sutton said. Material is reviewed twice - once on site, where spotters immediately halt the progress of backhoes if they glimpse anything resembling human remains. That material goes to a retired landfill on Staten Island, where it's sifted on a conveyor belt with people looking on both sides.
If deemed an official sample, it's shipped to the morgue, where an anthropologist determines whether it's human and whether it's from one person.
Human remains go via body bags to a temporary morgue set up by the Disaster Mortuary Operational Response Team. However, before a sample leaves, a small portion is shipped to the Forensic Biology Department.
"That's where we come in," Cash said.
The remains are matched against one or more of the following: existing tissue samples from victims, which is most reliable; DNA collected from possessions such as a toothbrush or comb; or DNA collected from relatives.
There are three major hurdles to identification.
First, some remains aren't there to be found; they disappeared in the fire and were pulverized under the weight of the towers.
Second, some remains are too badly burned to yield identification.
Third, tainted comparison samples caused by glitches in the chain of custody hinder proper identification.
Early confusion after Sept. 11 caused labeling missteps. For example, a comparison sample labeled "father" could mean the victim is the father, or it could mean the person who provided the sample to authorities is the father.
The digging should cease in about a month, and the forensics lab probably has six or eight weeks of testing after that.
More than half of the 17,000 remains tested haven't yielded good results, Shaler said. Phases two and three of testing use different technology but still use Gene Codes' software.