he two rival teams that decoded the human genome met on neutral ground last week to exchange notes on the merits of their respective approaches. The meeting, from which the teams' leaders were absent, was conducted by the computational biologists on each side in a cordial atmosphere that gave little hint of the tense and often acrid three-year race in which they had been involved.

But it would be premature to say peace had broken out. While the foot soldiers were fraternizing across the trenches, the generals on the two sides were engaged in a shoot-out in the scientific literature as to who deserves credit for cracking the human genome.

The contestants are a publicly financed consortium of academic centers and Celera Genomics of Rockville, Md. In a sudden truce last year, the two sides agreed to publish their interpretations of the human genome at the same time and to hold a joint conference to discuss discrepancies. The publications appeared together in February, though in rival scientific journals. Last week's meeting, the other element of the truce, was a deliberately low-key affair organized by the leading computational biologists on each side, Dr. Eugene W. Myers of Celera Genomics and Dr. David Haussler of the University of California at Santa Cruz. It was held at the villa of the Howard Hughes Medical Institute in Chevy Chase, Md., and it focused on general issues like the best way to decode mammalian genomes.

As it happened, Dr. Myers and Dr. Haussler were graduate students together in the early 1980's at the University of Colorado and were first exposed to the idea of applying mathematics to biology by their adviser, Dr. Andrzej Ehrenfeucht. Dr. Myers, in opening the meeting, said all the participants were friends and colleagues looking for common ground, and no speaker suggested otherwise.

As if by agreement to avoid divisive topics, no one directly mentioned the outstanding issue between the leaders of the two sides, a charge by Dr. Eric S. Lander of the Whitehead Institute that Celera's method for reconstructing the human genome was a "flop" and that the company could not have assembled its version of the genome without relying indirectly on the consortium's assembly method.

But a Celera scientist, Dr. Granger Sutton, reported that the company had recently assembled the human genome using only its own data, with slightly better results than its version of February, which used the public data as well.