An Earth-like atmosphere has a negligible mass compared to a planet, and "messing up the chemistry" is different from "completely removing the atmosphere".
GRB emit something like 10
44 J in a cone of a few degrees width, something like 1/3000 of the full solid angle. Values
from here. Inside the cone, the intensity corresponds to an isotropic emission of 3*10
47 J.
Earth has a gravitational binding energy of
2*1032 J.
A naive application of the inverse square law shows the planet doesn't receive enough energy to be disintegrated if it is more than
820 AU away. This value is much smaller for heavier planets (
90 AU for Jupiter) and larger for smaller planets.
A planet in the system of the exploding star could receive enough energy. There are two caveats, however:
- the energy is emitted in a narrow beam aligned with the stellar rotation axis. In most systems, this is outside the plane where the planets orbit. The intense main beam will miss the planets.
- The assumption that all the energy hitting the planet goes into disassembling it is unrealistic. You would get a scenario like
the last one here, where you quickly get a plasma layer and then most energy is absorbed by a surface layer of plasma.