|Using the web-based L2L tool||Making sense of L2L results|
|Using L2L from the command line||Batch mode from the command line|
Analyzing your microarray data with the web-based L2L Microarray Analysis Tool requires five simple steps:
1. Give your analysis a short name
The name should be fairly unique and meaningful (all of your results files will use this name), but must contain only letters, numbers, and underscores. No spaces or special characters, for security reasons. The web site will warn you if your name has any illegal characters, and will ask you to change it.
2. Upload your data file
Your data is a list of the probes that went either up or down in your microarray experiment. You can usually copy-and-paste the probe identifiers directly from whatever program you used to analyze your chips, but check file format help to make sure your probe identifiers look like what L2L is expecting for your microarray system. Make sure each probe is on a separate line, like so:
Use separate lists for each experimental condition, and for up- and down-regulated genes in that condition. For example, if your experiment involved two conditions (+drug1 and +drug2, both vs. no-drug control), you'll be analyzing four lists: drug1_up, drug1_down, drug2_up and drug2_down.
3. Select your microarray system
L2L has built-in support for a variety of popular microarray systems for four species: human, mouse, rat and rhesus macaque. Select the one you used from the pop-up menu. This tells L2L how to translate your probe identifiers to HUGO names and back again. If the system you used isn't there, you can create your own translator for whatever system you used. The "More Options" page will let you upload that translator in addition to your data.
4. Select a database for comparison
L2L compares your data to each of the lists in its database (aka "set of lists"), and looks for similarities. There are six sets of lists to choose from. The L2L Microarray Database (L2L MDB) is a compilation of published microarray results, and is the most novel and useful aspect of L2L. If you came here, you'll want to compare your data to the L2L MDB. There are also three sets of lists derived from Gene Ontology categories: Biological Process, Molecular Function and Cell Component. The L2L Microarray Analysis tool is a convenient way to see which of these Gene Ontology categories contain an over-abundance of your genes of interest (other tools, like EASE and GOMiner provide similar functionality). There are also sets of lists derived from predicted human microRNA targets; gene expression "modules" deduced from a large microarray study of 22 types of cancer; and protein-protein interactions from Reactome.
5. Press the button!
You'll be taken to a waiting-room page which will tell you about how long the anaylsis will take. Click on the link to view your results (what will my results look like?). If the results page isn't available yet, wait a few seconds and try again.
The output pages of L2L come in three flavors: Results Summary, Listmatch, and Probematch. The Results Summary page summarizes all of the lists in the database that overlapped significantly with your data. Each list in the database has a Listmatch page, which shows all of the gene in your data that match that list. Similarly, each probe in your data has a Probematch page that shows all of the lists that probe was found on. When you click on the link to view your results, you are taken to the Results Summary page. All pages have a navigation bar at the top that takes you back to the Results Summary, to a listing of all the lists your data was compared to (each linked to its Listmatch page), or to a listing of all the probes in your data (each linked to its Probematch page).
Interpreting the overlaps that L2L identifies can be challenging, but you may find it useful to look for clusters of lists that represent similar conditions, which all significantly overlap your data. This may be an indication that the condition represented by those lists is related to the condition that produced your data. Since most lists come in "up" and "down" flavors, as does (probably) your data, look for lists whose "up" flavor overlaps with one of your data files, and whose "down' flavor overlaps with the other - a strong sign that the stimulus that caused the differentially expressed genes on the database lists is also at work in your experiment. You may then want to see which genes in your data are responsible for the most interesting and significant overlaps (on the Listmatch pages), and perhaps look at which other conditions regulate the expression of those genes (on the Genematch pages).
If you have questions about the statistics behind L2L, or how to interpret the P values it generates, you can find more information on the statistics page. We are also happy to provide collaborative assistance in interpreting and validating L2L results.
1. Results Summary (sample)
The Results Summary page has a table with all of the lists in the database that overlapped your data with a p-value of less than 0.01. From left to right, the columns of the table show:
- the name of the list, linked to its annotation (PubMed abstract or AmiGO page)
- a description of the list
- the total number of probes on your microarray that represented genes on the list
- the number of probes derived from the list that would match your data by random chance
- the actual number of matches to your data, linked to the Listmatch page for the list
- the fold-enrichment of matches to your data (actual/expected)
- p-value representing the statistical significance of the overlap, derived from a binomial distribution
2. Listmatch pages (sample)
You can view a Listmatch page for a particular database list either by clicking the "Actual matches" number on the Results Summary page, or by clicking the list's name on the "View all lists..." page (sample). Each Listmatch page has a table showing all of the probes from your data that were found on that list. The probes are annotated in a variety of ways, from left to right:
- probe ID, linked to the Probematch page for that probe
- HUGO gene symbol
- color-coded at-a-glance functional annotations, derived from Gene Ontology associations
- links to the Gene Cards and Entrez Gene entries for that gene - HUGO gene name (description)
3. Probematch pages (sample)
You can view the Probematch page for a particular probe in your data either by clicking the probe ID on any Listmatch page, or by clicking the probe ID on the "View all genes..." page (sample). Each Probematch page summarizes all of the annotations for that probe (the same information presented on the Listmatch pages), then shows all of the lists in the database on which that gene was found. Click on any list name to view the Listmatch page for that list.
Once you have downloaded and installed L2L, open your terminal application and "cd" into the L2L directory. Launch L2L with "./l2l". You will be presented with a simple textual interface that prompts you for the location of the three necessary files: the data to be analyzed, the translator file to use, and the set of lists against which the data will be analyzed. The textual interface only permits a simple analysis, like the website - one data file, using one translator library, against one set of lists. However, you can customize any of these components however you like, and tell L2L to use your customized files. This may be especially useful for using a custom set of lists.
Once the analysis is finished, L2L will quit. A new folder will have been created wherever your data file resided. Open the "YOURDATA_output.html" file in that folder to start browsing your results. For more details, please see the README file.
If L2L produces errors on the command line, and you aren't sure why, check these common sources of problems:
- Line endings. Make sure all of your text files (data, translator library, list) have UNIX line endings, not DOS or Mac line endings. You can change the line endings in most text editors.
- Spaces in file paths. L2L does not work properly if any of the filepaths you give it contain spaces. Make sure that your filenames do not contain spaces, but also check the folders and subfolders that those files are contained in.
The batch mode of L2L, invoked by using the switch "-e" when launching L2L, allows you to run extremely powerful and complicated analyses with a single command. You can analyze any number of data files, each using a different translator library, against any number of sets of lists. All input to L2L is passed as command-line switches and arguments. These are all described in detail in the README file. but a few of the most useful commands are:
- "./l2l -abef datafolder": The most powerful batch-processing command. The directory "datafolder" can contain any number of data files (-b), all of which must have a line in them ("#LIB<tab>platform") specifying the translator library to be used (-f). All of the data files will be analyzed against all six default L2L sets of lists (-a). A separate folder will be created for the results of each data file / set of lists combination.
-"./l2l -bcef data lists": Same as above, but all of the data in "data" will be processed against all of the sets of lists in "lists". The directory "lists" must contain one or more subdirectories of L2L list files (-c). All of the subdirectory names should be short, without spaces or strange characters. This is intended for users who create a custom set of lists, and want to be able to analyze their data against their custom set as well as some or all of the default sets simultaneously.
- "./l2l -e data/mydata library/u133 lists/l2lmdb": The simplest "-e" command, which runs only a simple analysis of the sort you could run from either the website or the textual interface. "mydata" must be a text file with a list of probe IDs from the Affy U133 microarray; the data will be processed against the L2L MDB.
- "./l2l -e -n l2loutput data/mydata library/u133 lists/l2lmdb": Same as above, but the output files will be named "l2loutput" (-n) instead of "mydata".
In addition to detailed documentation of the arguments and switches, README file also contains a number of experimental scenarios in which you might find batch mode useful, with an explanation of how to take advantage of it.back to top