facs remove -o

[henrikstranneheim]

When using the -o option: -o /proj/b2012037/private/henrik/Example/Example1

yields: /proj/b2012037/private/henrik/Example/Example1Example1_Homo_sapiens.GRCh37.70_nochr_clean.fastq

making it look more like "-o" is an out directory command. When specifying '-o' then the suffix "Bloom_filter.clean" or "Bloom_filter.contaminants" should be added to the specified file name (if such exists). --> Example1_Homo_sapiens_GRCh37.clean & Example1_Homo_sapiens_GRCh37.contaminants. If "-o" path ends in a "/" indicating a directory then the complete entry as in the first paragraph should be used.

[brainstorm]

@henrikstranneheim @tzcoolman, @arvestad, I think the canonical way for any UNIX command regarding outputs is to exactly output the name that the user has specified:

-o whatever

Should generate "whatever" as an ouput file, nothing more, nothing less. Otherwise it's very confusing, IMHO.

[henrikstranneheim]

Yeah agreed. The trouble is that we are creating 2 files from 1. And if we are using a list we are creating 2 files for every Bloom filter specified. Here is a suggestion:

-od/--outdirectory --> Files are names according to: datasetName_BloomfilterName.clean.infileformat & datasetName_BloomfilterName.contaminants.infileformat

-fp/--fileprefix FileName (as specified from user)

-fs append .clean and .contaminants respectively to the "-fp" option. -fs takes to option where 1st is the cleaned file and 2nd is the contaminated reads, which should be stated clearly in the help text.

you cannot use -od and -fp together. You must use -fp and -fs together.

Thoughts?

[brainstorm]

In general I think that this functionality should be outside of facs, scripted away as an example in the documentation (for loop that goes through the generated bloom filters), i.e:

        for ref in os.listdir(self.reference):
            qry = test_fname.fastq
            bf = os.path.splitext(ref)[0]+".bloom"

            facs.remove(qry, bf)

Or way simpler in bash.

But yes, @henrikstranneheim, I agree with your specification if it gets implemented.

[tzcoolman]

@brainstorm it has actually been fixed since several months ago. I ll close it in the next PR

[brainstorm]

Actually, that's true, we agreed upon using stdout for clean sequences and stderr for contaminated ones... Please make sure this fact is documented in the facs remove help-like commands and that this functionality works as advertised (all flag combinations/edge cases (empty file, wrong format, etc...)) work.

Now that you mention your PR... can you please review and merge mine if correct? ;)