I was asking myself the same question because some of the wording used is quite suggestive to a layman reader like me. But after some Wikipedia my answer is: no

I was asking myself the same question because some of the wording used is quite suggestive to a layman reader like me. But after some Wikipedia my answer is: no, the actual virus carries its blueprint in RNA form that doesn't go through that transcription step where the errors occur. In JVM lingo, it's as if one was .class and the other was a single-class .jar and somehow certain byte sequences were less reliable during unzip than others. And as a consequence, those byte sequence are avoided by programs routinely distributed as .jar but used quite generously by those distributed as .class. Now they took some snippets from a codebase that is deeply rooted in the ".class" tradition and inserted it into a vector from the ".jar" side and the unzip errors happened to be rare enough to not light up during QA.