Quick Reference: Examples for the experienced user

This page is meant to act as a quick reference guide for those who are already familiar with hermie but may need a syntax reminder. Each set of examples begins with one in the general form. Any words inside of angle-brackets (<>) are a description of the words that should be supplied by the user. Any words between square-brackets ([]) are optional and do not need to be included (however, it is probably what is being illustrated in the example). Note that options may be specified in any order, but they must come between hermie and organism. First-time users may want to read the tutorial for a detailed walk-through.

Bare-bones usage

For a fully automated run, you only need to provide an organism and a search mode.

Any combination of organism and search mode can be used. The commands $ hermie -list-organisms and $ hermie -list-modes list what is available. See Choosing a search mode for more information.

Name the output directory

By default, the results are put into a directory named 'pipeline'. Not very informative. You can specify a different name for the output directory. This is particularly useful doing multiple runs on the same MS2 files. Suppose you had a set of MS2 files that you wanted to try searching with two different protein databases. You could use these two commands from the same directory and have the results put into different places.

Specify MS2 files

The automated version uses all of the MS2 files in the current working directory. For more flexibility, you can specify which MS2 files to run.

Instead of listing a specific file, you can give a directory and all files ending in 'ms2', 'cms2' or 'bms2' in that directory will be used. The current directory is specified with '.' (a dot). Directories and file names can be mixed together.

Use a different database

The organism defines the database (fasta file) used for the SEQUEST and/or crux search and the library used for the BiblioSpec search. By using the organism other and the -fasta option, you can specify a non-standard database. Suppose the database you want is in /share/database/platypus.fasta. You will also either need to specify a library or turn off the library search.

If you are running percolator, you also need to specify a decoy database. For this example, it is in the same location and named 'platypus-rand.fasta'.

Turn off steps

Sometimes you want to skip certain steps in the analysis. This is particularly useful if your run was interrupted and you want to finish the steps that were not completed. For instance, if your run was interrupted during the DTASelect step, you would repeat the same command adding options to turn off all of the preceding steps.

Add options for component programs

Hermie allows the user to pass options to the component programs. For example, you might want to change the threshold used by DTASelect after a percolator run. The default behaviour is to use a maximum FDR of 1%. In order to do this, hermie runs this command.

(The values are negated so that DTASelect will return values greater than -0.01, which actually means values less than 0.01) To use a less stringent threshold of 5% FDR, you would want DTASelect to use the option -S -0.05. The option-value pair will be written with an = in place of the space (i.e. -S=-0.05) You tell hermie which program to pass it to with the option --dta-sequest. The final hermie command will look like this.

It is possible to pass DTASelect (or any other component program) more than one option. Simply insert another --dta-sequest option=value in the command. For instance, you can ignore any proteins with the string "ribosome" in the name with the option -e ribosome. Adding it to the above command looks like this.

As another example, suppose you want BiblioSpec to search spectra of all charge states instead of limiting the candidates to those with the same charge as the query spectrum. The charge feature is controlled with the -z option. You would want hermie to run this command.

In this case, the option does not take an argument, so the option-value pair will be written as -z=. The final hermie command will look like this.

Using a custom mode file

Any of these options can be put into a separate file instead of being given on the command line. A file containing a collection of options is called a mode file (at least in the world of hermie). A detailed description of how to create and use custom mode files is given on the customization page. Here are the basics.

Let's say you put together a mode file and named it myfancysearch. Make sure the file is in your mode path. If your mode file does not contain options for naming the database or library, your command might look like this.

If you included a custom database in your mode file (options -fasta and -decoy), then you need to use the organism "other". If you use "other", you also must specify a library or turn off the library search option. Remember that the --noblib-search or -library options could alternatively be included in the mode file.

Copying results to the web

The results you are most likely interested in are in the DTASelect.html file. You can have this file copied to your web directory automatically with the -web option. If you provide a directory, the copied file will still be named DTASelect.html. If you provide a directory and file name, the file will be renamed. This option will over-write a file of the same name.

Email notification

Hermie runs can take days to complete. Wouldn't it be nice if you didn't have to keep checking back to see if it was done? Now you don't have to. You can have an email sent to you when the run has finished.

Running crux

There are now two different database search algorithms available in the hermie pipeline, SEQUEST and crux. SEQUEST is run by default. Crux can be run in addition to or instead of SEQUEST. When using pre-defined organisms, crux will use an indexed version of the standard fasta. For custom databases use the -fasta option. Crux generates its own decoy sequences on the fly, so no decoy database is necessary. Crux results will also be analyzed by percolator when that step is turned on.

All crux options should be put in a parameter file and the name of the parameter file given to hermie with --crux-option. The path to the parameter file will be passed to crux as is, so a fully-qualified path (one begining with '/') is safest. Note that hermie actually runs parallel-crux, a wrapper that divides the search into blocks of scans and parcels them out to different cluster nodes. However, the options you specify with --crux-option should be crux options. A few options are disabled when crux is run via parallel-crux. See the parallel-crux documentation for details.

A run including crux might look like this.