Luis Pedro Coelho

Over the long term, I would prefer an approach where, whenever `ngless` runs¸ it will create any missing outputs. The whole `lock1`/`collect` business is a bit of a hack now....

> Another perhaps simpler alternative is to add a stats_label="Human reads kept" argument to all functions that can generate statistics. I like this idea, it could be made more general,...

[biom-format](http://biom-format.org/) or netCDF seem good candidates. Either way, a binary format would be helpful for downstream processing.

+1 but this is hard to implement with the current structure. At the moment, the locks will eventually expire, but I agree it can be annoying to wait for this.

I agree that these could be interesting. In general a more comprehensive syntax for defining regions in `select`/`count` would be a good thing for ngless 2, but maybe an intermediate...

Could also be a flag when saving (to the `write` function). Not 100% sure what is the best model.

The `write` function already has a `format_flags` argument, so it could be `write(newdata, ofile="...", format_flags={no_qualities})`. @unode: what use-cases do you see with the second interface? I am not against it,...

In principle, we could move the stripping to earlier in the pipeline as an optimization later without changing the user-visible interface.

There is a hacky way to do this, which is to do `--subsample`. Not a great solution in terms of UX, but it works for now.

The error message is pretty awful, but I have seen this happen when BWA runs out of RAM and crashes in a bad way. The ocean demo requires a lot...