shaze / misc

misc

figsplit.py

This script is used to render PDF from xfig that is suitable for LaTeX Beamer presentations, where the xfig has multiple depths or layers. I used to generate multimetapost files and then use mpost to generate multiple PDF. However, recent versions of xfig seem to have dropped this.

In the simplest form run

figsplit.py picture.fig

This produces picture-0.pdf picture-1.pdf picture-2.pdf.... Then, use the xmpmulti package in LaTeX and use the multiinclude function (perhaps don't use the <+>)

  \multiinclude[<+>][format=pdf,graphics={width=9cm}]{picture}

Notice that items with lower depth are above items with higher depth. The default behaviour is that everything that at depth n also occurs in the PDFs for depth below n. So if depth 40 correponds to picture-4.pdf then all items at depth 40 will also occur in picture-5.pdf, picture-6... Usually this is what we want, but sometimes in a dynamic picture you will want avoid this

figsplit.py --occlude 30-40:40.41,20-25:28 picture.fig

This says -- in the PDFs produced in layers 30-40, don't include items at depth 40 or 41, and in layers 20-25 don't include items at depth 28

parallel.nf

Run like this

nextflow run parallel.nf --script scriptfile 

scriptfile should be a bash file with a separate command on each line -- the commands should be run in the foreground not the backgroud. There are two examples given sleep.cmd and plink.cmd

Each command will be run in parallel on the system, subject to the maximum capacity of your system. The following options are permitted

• -profile slurm submit each job separately to the head node of the cluster running slurm. NB. This parameter takes a single "-" Note that you need to have the nextflow.config file in your directory when you run it. You may need to change the queue name to which your jobs should be submitted
• -profile pbs ditto for a cluster running PBS
• --cpus show many CPUs does each job need. The default value is 1.
• --mem how much memory is required. Default is "1GB". The unit should be specified
• --max_fork how many jobs should Nextflow allow to be active at a time (default is 30, probably 60-100 would be fine but that would depend on the size of your cluster, how busy it is, and any per-user limits on your job. You should generally rely on the scheduler to do its job but be aware of others needs -- definitely if you have a thousands of jobs or if each job is very long running, pick a reaonable number not to antagonise your colleagues.)
• --out_dir the name of the directory where you want the result to go

For example

nextflow run parallel.nf --script plink.cmd -profile slurm --mem 6GB  --cpus 2 --max_fork 20 --out_dir /tmp/result

Note that if the number of jobs is much larger than max_forks or if the jobs are very long running, you should run a screen session for the nextflow run.

collect.py

Remote fetching of data.

Here are instructions that will create a folder in your home directory with the necessary set up In the terminal sessions

1. Open up a Terminal session (on a Mac, you can find Terminal in the Utilities folder of your Applications folder).
2. Type git clone https://github.com/shaze/misc.git This will create a directory called misc in your home directory. If you already have a file or folder with that name say git https://github.com/shaze/misc.git mymisc and then do the following instructions mutatis mutandis using *mymisc" ,
3. Change directory: On my system this would be cd misc
4. If you type in ls (followed by return) you should several files including sample.csv and collect.py
5. Run the program. In the example below I am assuming your file is called sample.csv. Replace this whatever you have called your data file. Type in the command python collect.py sample.csv followed by return.
6. The program tries to guess the correct ouput type but can't always do that.
7. If the program can't fetch the data : e.g., file no longer available you should get a meaningful error message