The Caltech researchers discovered the rise in B.1.526 by scanning for mutations in hundreds of thousands of viral genetic sequences in a database called GISAID. “There was a pattern that was recurring, and a group of isolates concentrated in the New York region that I hadn’t seen,” said Anthony West, a computational biologist at Caltech.

He and his colleagues found two versions of the coronavirus increasing in frequency: one with the E484K mutation seen in South Africa and Brazil, which is thought to help the virus partially dodge the vaccines; and another with a mutation called S477N, which may affect how tightly the virus binds to human cells.

By mid-February, the two together accounted for about 27 percent of New York City viral sequences deposited into the database, Dr. West said. (For the moment, both are grouped together as B.1.526.)