biological information flow
Biology is, at bottom, an information business: the precise order of building blocks in one molecule gets used as a pattern to set the order of building blocks in another. Biological information flow is the study of which of those copying steps actually happen, in which direction, and how faithfully. The 'information' here is concrete, not metaphorical: it is the linear sequence of letters along DNA and RNA, and the linear sequence of amino acids along a protein.
Crick laid out the possible transfers as a small grid. The transfers that happen routinely in normal cells are DNA to DNA (replication, when a cell divides), DNA to RNA (transcription), and RNA to protein (translation). Some transfers happen only in special cases: RNA to DNA (reverse transcription, in retroviruses and in our own telomerase and retrotransposons) and RNA to RNA (replication of some RNA viruses). And some transfers appear never to happen at all in nature: protein to protein, protein to RNA, and protein to DNA. The forbidden ones all share a feature: they would require reading sequence information out of a protein.
Keeping this grid straight is what stops common confusions. A protein can absolutely influence which genes are switched on, but that is regulation, not information transfer in Crick's sense, because the protein does not dictate the actual letter sequence of any gene. Prions, infectious misfolded proteins, propagate a shape rather than a sequence, so they too leave the rule intact. Thinking clearly about information flow tells you what is genuinely surprising in biology and what merely looks surprising.
Telomerase, the enzyme that maintains the ends of our chromosomes, carries a short RNA template and uses it to add DNA. That is an RNA-to-DNA transfer happening inside healthy human cells, not just in viruses.
Even our own cells run information 'backward' from RNA to DNA.
Regulation is not information transfer. A transcription factor protein deciding whether a gene is read does not copy any sequence; it only opens or closes a switch. Only sequence-copying steps count in Crick's grid.