| |
|
|
|
|
|
| |
|
|
|
|
| Q. What is GONOME ? |
| A. GONOME is an application for finding significantly over- or under-represented Gene Ontology terms associated to genomic positions. While there are a number of tools that deliver this type of statistics for micro-array data, they consider all genes as equally probable, however for genomic positions the varying sizes of genes means that a random position is more likely to fall into a larger gene than a smaller one. GONOME calculates over-represented GO terms taking gene length into account. GONOME takes a set of genomic positions, finds the gene into which each position falls, or the upstream and downstream genes in the case of intergenic positions, sums up the number of times each GO term occurs and then calculates how significantly thatsum is over-represented compared to where random positions would be expected to fall. Full details of the method are available in the paper. |
| Q. What are the requirements for running the program locally ? |
| A. You need perl and the ability to install a few CPAN modules. A C++ compiler would be handy, however there are precompiled versions of the C++ component, multiprob, for windows and linux i386. More details are available in the README of the package available in the resources section. |
| Q. Is there a paper ? How do I reference this ? |
| A. Yes, currently heading for publication. Check back here for reference details. |
| Q. I get 'Ooops, can't get cookie value, are cookies enabled ???' What's wrong. |
| A. GONOME uses cookies to identify users, either cookies are disabled in your browser or your cookie has timed out, moral of the story : save your results locally, GONOME makes no promise they'll still be there tomorrow! |
| Q. What does GONOME consider 'significantly over-represented'? |
| A. An E-Value (expected times this term scores this high) < 0.05, or an expected number of occurences less than 1 time in 20, the traditional 5% confidence level. |
| Q. What's with the 'Z score to consider Over-represented' field ? Can I lose significant terms this way. |
| A. To reduce response time GONOME prefilters querys by not calculating probabilities for terms which cannot be significant being less than X, the chosen parameter, standard deviations away from the mean. 3 is a safe value in all cases, 4 is generally safe for the larger GO splits, e.g. human process. Higher values can be used to cull higher expectation results and for a quicker first pass analysis with complicated queries. Further details, including a justification for the statement that no signficant terms can be lost with a cutoff of 3 are available here |
| Q. What's the 'Accuracy Threshold' ? |
| A. This corresponds to the number of significant figures to which the probability calculations are accurate. |
| Q. How can I find out which genes were associated to my term of interest ? |
| A. Tick the 'Add loci of hits to table' checkbox, and then look in the ActualVsExpected Table. |
| Q. I get 'No Positions Found' or no results in my output table, what's gone wrong ? |
| A. There are any number of possibilities I fear. Check that your input file contains positions, that it is in one of the accepted formats and has either contig or chromosome positions appropriate to the organism. If all looks good, raise a bug report above. |
| Q. What are the fields in the *.tab output files ? |
| A.
The .tab files contain the lengths necessary for multinomial
probability calculation by multiprob. These are tab delimited, the
first field is the GO term, which may also contain the number of hits
at the number of loci in brackets. This is useful for seeing clustered (multiple to one gene) hits. The second field is the actual number of hits to that term from the query positions set. The third and fourth are expected number of hits to the selected regions and its standard deviation. The remaining fields are ranges associated to the term in the first field. These are (fields 5-9) the transcribed, upstream, downstream and both ranges associated to the term. Both handles the case where a position matches the same term up and down and is within upstream and downstream cutoffs. |
| Q. Why do graph outputs show expectations however there is also a probabilities file, what's the difference ? |
| A. Simply, the expectation are the probabilities after apply Bonferroni correction, |