Set minimum chunk size for adaptive sequencing

I would like to perform adaptive sequencing ("Read Until") on a dataset with relatively short reads. As a result, if I want to gain significant benefit from selective sequencing, reads need to be ejected as soon as possible. In order to do so, I would like to reduce MinKnow's default read chunk size.

According to the comments in this repository, the configuration is stored in minknow/conf/app_conf. This is incorrect and should be updated; this information is now stored in minknow/conf/tuning_params.toml.

Would you additionally be able to explain the difference between the following configuration options in minknow/conf/tuning_params.toml:

• raw_data_intermediate
• raw_meta_data_intermediate
• read_data_intermediate
• event_data_intermediate

Are there any limitations I should be aware of in terms of how low I can set these parameters? I assume once the data packets are too fragmented it becomes difficult to write them in real time. It looks like the default values are already significantly lower than they used to be (400 samples, not 2000). Is that correct?

Start with the configuration comment in minknow_api/data.proto and compare it with minknow/conf/tuning_params.toml. Trace the four named parameters and their documented defaults or limits, then update the configuration location and explain their differences and constraints in the issue or relevant documentation.