NGLess 1.5 Ideas

[Luis Pedro Coelho]

As I read through the open issues on github (https://github.com/ngless-toolkit/ngless/issues), I see three clusters that dominate and I have some ideas to work them out

1. The parallel module

I have some ideas to simplify both the API and the internals

2. The stats computation

This is currently somewhat messy, especially as it interacts with the parallel module (I actually think it works well for single samples, but is not great for multi-sample). My current thinking is to replace this with an IPython notebook.

3. Importing local modules

This may require a bit more thinking, but maybe just some better messages when someone tries to import a non-existent module.

Best,

Luis

Luis Pedro Coelho | Fudan University | https://luispedro.org

https://orcid.org/0000-0002-9280-7885

[ullo...@googlemail.com]

Dear Luis,

not directly related to ngless, but the ngless contributions: a metaphlan update to version 3 maybe? And would it make sense to include a rarefaction tool e.g. RTK? Also, would it maybe be a good idea to include creating a ngless.conf file with defaults during the install process? For less advanced users later it's easier to modify than building it from scratch I guess. Since gmgc is now an internal module, would it be possible to fix the "long headers" issue (I know this is somehow not related to ngless too, but somehow it is) or write a warning somewhere that samtools >v1.6 can cause errors (or even catch errors from samtools and guide the user in this case via stderr?). I will think about some more valuable feedback the next days.

Thanks a lot!

Ulrike