search

NCBI-Hackathons / Network_SNPs

A framework for network analysis and display of SNPs
Creative Commons Zero v1.0 Universal
17 stars 5 forks source link

Reorganize directories #6

Open RadixSeven opened 8 years ago

RadixSeven commented 8 years ago

Directory structure is crazy. We need to agree on a structure and move things around. 

find ~/ubuntu -type f -print0 | xargs -0 grep -l old_dir_name

Should give an idea of which files have dependencies on the old structure. (note that the command above hasn't been actually run and may contain typos)

graphanalytics commented 8 years ago

I totally agree. Would you mind proposing a structure that is optimal in your opinion ?

RadixSeven commented 8 years ago

I can't propose optional, but I think an improvement would be:

~/analyzer/
~/analyzer/hotnet2
~/analyzer/funseq
~/analyzer/better_name_for_networkx
~/analyzer/dmGWAS
~/analyzer/..........
~/meta-analysis -- where converters and meta-net-var.py live
~/networks
~/sample_data
~/docs

Is this missing anything you can see?

graphanalytics commented 8 years ago

This is great. I would not change a thing in your proposed structure. could we rename funseq to funseq2 ? I propose - simplenetvar - as a better name for oscar's module - as it basically does the simple/straightforward/generic network analysis of variants. Wonder how this structure could be represented in the github page. Team, please do exchange your thoughts on this.

DCGenomics commented 8 years ago

SGTM!

On Thu, Jan 7, 2016 at 12:08 PM, graphanalytics notifications@github.com wrote:

This is great. I would not change a thing in your proposed structure. could we rename funseq to funseq2 ? I propose - simplenetvar - as a better name for oscar's module - as it basically does the simple/straightforward/generic network analysis of variants. Wonder how this structure could be represented in the github page. Team, please do exchange your thoughts on this.

— Reply to this email directly or view it on GitHub https://github.com/NCBI-Hackathons/Network_SNPs/issues/6#issuecomment-169732305 .

What have you done today to make the world a better place?

RadixSeven commented 8 years ago

As a tool for representing the directory structure, I just ran across:

http://www.gravizo.com/

which is tailor-made for our purpose:

![Directory structure](http://g.gravizo.com/g? digraph G { home -> analyzer; home -> meta_analysis; meta_analysis [label="meta-analysis"]; home -> networks; home -> sample_data; home -> docs; analyzer -> hotnet2; analyzer -> funseq2; analyzer -> simplenetvar; analyzer -> dmgwas; analyzer -> more; })

The source for the above is:

![Directory structure](http://g.gravizo.com/g?
 digraph G {
   home -> analyzer;
   home -> meta_analysis;
   meta_analysis [label="meta-analysis"];
   home -> networks;
   home -> sample_data;
   home -> docs;
   analyzer -> hotnet2;
   analyzer -> funseq2;
   analyzer -> simplenetvar;
   analyzer -> dmgwas;
   analyzer -> more;
 }
)
graphanalytics commented 8 years ago

Looks cool. Thanks for sharing this.

hagenaue commented 8 years ago

Beautiful! Should we create a data input and output folder too?

On Thu, Jan 7, 2016 at 12:44 PM, graphanalytics notifications@github.com wrote:

Looks cool. Thanks for sharing this.

— Reply to this email directly or view it on GitHub https://github.com/NCBI-Hackathons/Network_SNPs/issues/6#issuecomment-169752284 .

Megan Hastings Hagenauer, Ph.D.

Research Investigator in the Molecular, Behavioral Neuroscience Institute Instructor in the Department of Psychology University of Michigan-Ann Arbor

RadixSeven commented 8 years ago

@hagenaue I don't think so. Besides test inputs, we are not providing any inputs. The user should feel free to put their inputs and outputs where they like. If I were the user, I'd probably structure things by analysis like this:

![Directory structure](http://g.gravizo.com/g? digraph G { home -> cure_cancer; home -> cure_cystic_fibrosis; cure_cancer->ca_networks; cure_cancer->ca_snps; cure_cancer->ca_outputs; cure_cystic_fibrosis->cf_networks; cure_cystic_fibrosis->cf_snps; cure_cystic_fibrosis->cf_outputs; cf_networks [label=networks]; cf_snps [label=snps]; cf_outputs [label=outputs]; ca_networks [label=networks]; ca_snps [label=snps]; ca_outputs [label=outputs]; })

That way, all of my data for my two nobel-prize-worthy publications would all be in one place.

However other users might prefer to structure it by file-type as you suggested: ![Directory structure](http://g.gravizo.com/g? digraph G { home -> networks -> n_cure_cancer; home -> snps -> s_cure_cancer; home -> outputs -> o_cure_cancer; networks -> n_cure_cystic_fibrosis; snps -> s_cure_cystic_fibrosis; outputs -> o_cure_cystic_fibrosis; n_cure_cancer [label=cure_cancer] s_cure_cancer [label=cure_cancer] o_cure_cancer [label=cure_cancer] n_cure_cystic_fibrosis [label=cure_cystic_fibrosis] s_cure_cystic_fibrosis [label=cure_cystic_fibrosis] o_cure_cystic_fibrosis [label=cure_cystic_fibrosis] })

I think that each of these ways are equally valid and we should let the user choose rather than encoding one of them directly in the repository.

graphanalytics commented 8 years ago

should the root of the structure be /home/ubuntu/ rather than /home/ ? because /home/ would be the system level directory in unix... i created analyzer withing /home/ubuntu/ ..... does that make sense ?

nyetsche commented 8 years ago

Yes, it should be /home/ubuntu, but actually, I'd suggest even rewriting it to be $HOME/Network_SNPs, (or a pure relative directory), so the Github checkout works even when not using AWS.

hagenaue commented 8 years ago

/home/ubuntu didn't seem to work, so I have left it with the default, which is outputting to the vegas folder. Since it is only output from a sub-step in the pipeline (not final output), I figure it is probably not worth picking at anymore.