Preparing NGLess 1.3
24 views
Skip to first unread message
Luis Pedro Coelho
Jan 20, 2021, 12:20:44 AM1/20/21
to NGLess List
Hi everyone,
I think we could do NGLess 1.3 soon. Like for 1.2, there haven't been any big "killer features" changes, but a bunch of small improvements (maybe most notably, read_int() and read_double() functions + to_string() methods).
If there are any small things that you think could be done, now is a good time to ask for them.
Best
Luis
Luis Pedro Coelho | Fudan University | http://luispedro.org
https://orcid.org/0000-0002-9280-7885
I think we could do NGLess 1.3 soon. Like for 1.2, there haven't been any big "killer features" changes, but a bunch of small improvements (maybe most notably, read_int() and read_double() functions + to_string() methods).
If there are any small things that you think could be done, now is a good time to ask for them.
Best
Luis
Luis Pedro Coelho | Fudan University | http://luispedro.org
https://orcid.org/0000-0002-9280-7885
Vadim Puller
Jan 22, 2021, 3:55:46 AM1/22/21
to NGLess
Hi Luis,
A minor suggestion: including feature length in the count table.
In a bit more details: `count` function may output the results normalized by the feature lengths: `{raw/normed/scaled}`. From the point of view of the downstream data manipulation, I would prefer to have raw reads that I could normalize by the feature length when necessary. As of now I have to import these lengths from the catalogue and merge them with the count table outside of NGLess.
A less minor disadvantage is the necessity to launch NGLess as many times as there are the samples. Thus, one has to use an external script or a different software for organizing workflow, reducing NGLess to a command line tool. But I suppose this is not counted as "small improvements"
Regards, Vadim.
A minor suggestion: including feature length in the count table.
In a bit more details: `count` function may output the results normalized by the feature lengths: `{raw/normed/scaled}`. From the point of view of the downstream data manipulation, I would prefer to have raw reads that I could normalize by the feature length when necessary. As of now I have to import these lengths from the catalogue and merge them with the count table outside of NGLess.
A less minor disadvantage is the necessity to launch NGLess as many times as there are the samples. Thus, one has to use an external script or a different software for organizing workflow, reducing NGLess to a command line tool. But I suppose this is not counted as "small improvements"
Regards, Vadim.
Ulrike Löber
Jan 22, 2021, 4:01:07 AM1/22/21
to NGLess, Vadim Puller
Dear Luis,
I agree with Vadim regarding the parallel feature.
Could it be that the syntax for different catalogues differ? e.g. KEGG_KOs for IGC, KEGG_ko for GMGC? Could this be harmonized by setting aliasses at least for "built in" modules?
Best,
Ulrike
Am Freitag, 22. Januar 2021, 09:55:48 MEZ hat Vadim Puller <vpu...@gmail.com> Folgendes geschrieben:
--
You received this message because you are subscribed to the Google Groups "NGLess" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ngless+un...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/ngless/00a73c57-eebe-4bf7-9fff-7320619bce1dn%40googlegroups.comYou received this message because you are subscribed to the Google Groups "NGLess" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ngless+un...@googlegroups.com.
To view this discussion on the web visit
.
Luis Pedro Coelho
Jan 22, 2021, 4:44:32 AM1/22/21
to NGLess List
1. I actually agree also that the parallel module's API is suboptimal. Maybe this is something that we improve for 1.4, so that there is a longer test period, though. I think a real solution may have to wait a bit longer still, but one simple improvement would be for NGLess to restart itself every time it is not done (this would be more similar to how Jug works, which was the original inspiration for this way of working).
2. We are implicitly becoming more consistent as we use eggnog-mapper2 for everything, but I agree that it has been done a bit haphazardly before.
HTH,
To view this discussion on the web visit https://groups.google.com/d/msgid/ngless/2119313875.1385396.1611306037361%40mail.yahoo.com.
Ulrike Löber
Jan 22, 2021, 4:49:51 AM1/22/21
to Luis Pedro Coelho, NGLess List
Dear Luis,
This might be even an issue which somehow affects --validate-only which does not check the feature headers of the mapping files I guess?
Results in positive scrip validation and abort process due to invalid syntax.
Best,
Ulrike
Am Fr., Jan. 22, 2021 at 10:44 schrieb Luis Pedro Coelho
To view this discussion on the web visit https://groups.google.com/d/msgid/ngless/d81e3751-0bb3-4126-bb0c-fa9b8cf9dc2d%40www.fastmail.com.
Luis Pedro Coelho
Jan 25, 2021, 5:37:24 AM1/25/21
to NGLess List
Oh, this is a very good point. The checks for headers are indeed not run on --validate-only (although they are run at the start of execution).
This does qualify as a simple enough fix that I can still make it for 1.3. It's just a matter of reorganizing the existing, code.
Tx
Reply all
Reply to author
Forward
0 new messages