FALCON assembler without SGE?

[Nordic]

Has anyone had any success getting FALCON to work on a single node without SGE installed?

-Jason

[rlleras]

Hello Jason,

Recently there has been an addition to the FALCON code base that allows it to run in stand alone more (via BASH). To do this, please wipe the SGE options empty, but do not remove those parameters, otherwise you will encounter an error. Also, you'll need to add the line 'job_type= local' to the CFG file. Lastly, make sure you are running the latest version to use this functionality.

Happy assembling!
- Roberto

[Nordic]

That worked Roberto, thank you very much!

[LanceLycan]

But even when I modified the main script (vim fc_run.py) into:

def run_script(job_data, job_type = "local" ):

And empty those params in fc_run.py:

sge_option_da =
sge_option_la =
sge_option_pda =
sge_option_pla =
sge_option_fc =
sge_option_cns =

I still get this error msg:

/export/arrayPRO2/PacBio/FALCON/fc_env/lib/python2.6/site-packages/falcon_kit-0.2.1-py2.6-linux-x86_64.egg/falcon_kit/FastaReader.py:40: DeprecationWarning: the md5 module is deprecated; use hashlib instead
import md5
sh: qsub: command not found

What do I do wrong?

[ATϟGC]

Hi Lance,

Please note: I am a beginner when it comes to FALCON and bioinformatics so what I did below might not be correct but I thought that I would share what I did with a similar error message.


I had a similar qsub error after installing FALCON and trying the E.coli example locally from a terminal in Ubuntu. It was unclear to me from rlleras' comment above which file had to have the options removed while retaing the parameters.
So I tried inserting 'job_type = local' into the fc_run_ecoli.cfg file above the first sge_option line so that it now reads:


job_type = local

sge_option_da = -pe smp 8 -q jobqueue


I then saved it as fc_run_ecoli_local.cfg and ran it with the following command:

path/to/my/FALCON-master/ecoli_test$ fc_run.py fc_run_ecoli_local.cfg


I think that FALCON ran properly and generated the correct files.

From what I understand from the FALCON manual, contigs were not generated from the above process so I tried constructing graph from overlaps like this:


path/to/my/FALCON-master/ecoli_test$ fc_ovlp_to_graph.py /path/to/my/FALCON-master/ecoli_test/2-asm-falcon/preads.ovl


Then I tried constructing the contigs from graph as follows :

path/to/my/FALCON-master/ecoli_test$ fc_graph_to_contig.py


note : Before entering the command above, I copied the 'preads4falcon shortcut' file from the 2-asm-falcon folder and pasted it into the 'e-coli_test' folder as this script seemed to be looking for it.


My question now is where do I look for the contig files and what are they called? Or did I do something incorrectly?

[ATϟGC]

Hi all,

I think that I have answered my question:

It appears that the E-coli genome contig (~4.6Mbp) is p_ctg.fa and is output in : path/to/my/FALCON-master/ecoli_test/2-asm-falcon

It appears that my extra steps of constructing graph from overlaps
and constructing the contigs from graph was unnecessary as these steps were already completed.

[LanceLycan]

Dear ATㄣＧＣ
Thank you so much!!! You made the program finally running on a none-cluster server.

[KamilSJaron]

Hello guys,

I am very happy to see this thread. How do you specify a number of concurrent jobs on local computer / number of cores / memory falcon can use at most???

I mean, I see that specification for resources are written on every job separately, but since it is not clear, how many jobs can run simultaneously, it I am not sure how to estimate it (and i do not want to overload our server).

If there is only one job at the time (which is kind of logical for all non-overlap jobs), I should probably specify parameter -sXXX of pa_DBsplit_option to something really big (to have only one process only).

I could probably find out by running it on lambda phage or so, but if you have done it already, I would appreciate, if you will share your experience...

Cheers,
Kamil