Add plots for the original FACS 1.0 dataset

[brainstorm]

We should get some metrics (runtime/accuracy) from the old dataset (the one used in the original paper) with the current version of FACS.

Preferrably in an automated/reproducible way. Manually if it is too cumbersome.

@tzcoolman, @henrikstranneheim, can you take care of that?

[tzcoolman]

@brainstorm I thought I had done that before. Plus, fastq screen only takes datasets with fastq format while the old one is in fasta. FACS2.0 and deconseq has slightly runtime difference in handling fasta and fastq format file.

[brainstorm]

Are those plots relevant today given all the changes introduced to FACS since then?

Can they be regenerated/made by anyone else than you and/or Henrik? Where is the code to do so?

Thanks Enze!

[tzcoolman]

It is reproducible of course, though I didn't use a script to run it. All I have is relevant data. (Same as what Henrik had done in the paper, only counting ecoli and human chr 8 and 22) @brainstorm

[tzcoolman]

@arvestad @guillermo-carrasco @brainstorm I rebuild a synthetic dataset which contains exactly the same compound (species types) and similar proportion as the old one (Henrik's) using simNGS. It can be easily merged into the std python testing module (ecoli ref test). Should I just directly send it to you or I should do the test?

[guillermo-carrasco]

Hej!

If you generated it with SimNGS it should be easily reproducible right? We just need the generation parameters to get the same output, no need to send lots of GB through the network!

[tzcoolman]

@guillermo-carrasco I just dont understand. All I used is default setting. Do you mean all the ref genomes that Henrik used before

[brainstorm]

@tzcoolman You should write a python test that:

1) Builds the old dataset using the "default setting" you mention. 2) Runs FACS against it. 3) Generate data points, or even plots out of it.

[guillermo-carrasco]

what I meant is tha you don't have to send the dataset, but a "Howto" generate it. Automating the procedure as @brainstorm suggests would be the best solution, of course.

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[tzcoolman]

@brainstorm It cannot be fully automatic. At least for 20 of the ref genomes, I dont know where to download them automatically

[brainstorm]

Alright, how big are those? Can you put them on your public dropbox account for now? I’ll figure out a better location for them.

[tzcoolman]

@brainstorm 64M without human chromosome ref. I guess for human HG19 chr 8 and chr21, it will be easy to automatically download it.

[tzcoolman]

@brainstorm.. I ll leave it on lars' desktop 'turing'. I haven't being able to use dropbox since June. It's about 100MB unzipped.