There are two common possibilities in real time.
1. You would have to reset the SEQ number at the Informatica level (in the transformation) to match the numbers at the database. This happens when the tables (dimensions or facts) are being updated by multiple processes. It’s always better to go with a single process for creating the unique SEQ numbers. Yes, there will be business cases, where there will be a need to have multiple processes updating the same table in those cases, you can have the SEQ as the first step (in the first process) and then the rest of the processes using the numbers created by the first process.
2. Database Updates. This happens when there are business requests to revert the load or if there is any wrong data loaded. Based on the project and the procedures on how data should be handled in production it is either manual process or thru an automated process (Should be setup using the ETL process).
In either of the cases, we have to make sure (via deployments or migration) that the SEQ’s are the same in all the environments to have proper data visibility and lineage.
** Please note there are multiple other situation and scariness which are environment/business specific and those should be considered before going with the above solutions**