Re: [GenMAPP] Re: Bugs in GeneMAPP

[Alexander Pico]

Hi Josep,

This is a common problem for identifier mapping. Just to describe the process in a bit more detail, so you can see what’s happening. Your dataset has HGNC ID (say, “ABC1”) and the pathway has a node labeled ABC1. You’d expect these to match! However, the pathway is only using ABC1 as a label and might be using a proper ID system, e.g., Entrez, to encode the node. If your dataset does not include the same type of identifier that is used in the pathway, then we have to perform ID mapping. So, your HGNC ID gets translated to Ensembl and then that gets translated to Entrez, which then maps (or doesn’t map) to nodes on the pathway.

This is an imperfect solution, but it is one of the most robust around. The only work around is to find out the underlying IDs used in the pathway and then key your datasets with those same IDs directly. Then you can skip the (imperfect) ID mapping process altogether. That’s a bit of work and exactly why we provide the ID mapping solution.

Hope this helps explain why some mappings don’t work. And thanks for the great feedback on the system!
 - Alex


On 7/19/12 4:38 AM, "Josep" <asime...@gmail.com> wrote:

Dear Nathan

Thanks for the invaluable information.

Just  say that we are working with custom probes, so I guess that no probe ID will be recognized. Therefore, I  prefer to work with gene names (we use the human nomenclature for the genes). When doing this, the main problem is that some genes - with exactly the same name in both my list and the loaded human network (from wikipathways) -  remain in white, like if they had not been recognised or no FC values had been assigned. The main part of genes, though, show the assigned colour correctly.

Just to let you know...

Thanks so much!
 

On Wednesday, July 18, 2012 6:58:38 PM UTC+2, Nathan Salomonis wrote:

Hi Josep,

It sounds like you are using GenMAPP-CS, correct?

See below:

1. When creating criteria, if I move the box where the colour is chosen, the whole thing disappears and only the buttons (Accept, cancel, etc) remain.

If so of the behaviors in #1 you described will be noted and examined further
 

I have imported a list of gene expression data. The list contains repeated gene symbols (different probes for the same gene). I was wondering:

Note, that GenMAPP accepts Affymetrix probeset IDs as identifiers, so using gene symbols as annotation is an unnecessary annotation step that may result in lost information.

2. how the software manages these replicas when looking for criterias.

When multiple distinct IDs in your dataset such as probesets map to the same gene, the values of each probeset are stored in the dataset, but when initially viewed on a pathway or network, the average of all the gene aligning probesets are shown for all numerical values. You can view the independent probeset results on the pathway if you change the metanode display options as described at the bottom of this page:
http://www.genmapp.org/beta/genmappcs/manual-viewpathway.html

If you have multiple lines in your spreadsheet for a gene and all have the same primary ID selected in GenMAPP-CS, however, only when line will be represented for those nodes. Hence, you must more descriptive IDs, such as probesets if you want to view data for each identifier.

3. Working with a network where one gene appears more than once in different positions (Nuclear Receptors in lipid metabolism and toxicity). The colour of the gene matches the criteria, but only in one position (ABCA1) . The rest of boxes where the gene name appears remain in white.

I was unable to replicate this problem. ABCA1 probesets map to the multiple instances of ABCA1 in the pathway fine in the dataset I am using.  This is a known use-case in GenMAPP-CS that should work fine.

4. I have seen that the software recognizes the most of the genes present in the list. But some genes, despite being in the original list (which is completely loaded, I can see it in the number of rows), seem to be not recognized and remain in white. This happens randomly, I have been unable to identify a common pattern.And the name of the gene in my list is identical to the name that appears in the network.

The name of the gene in the WikiPathway is not the actual ID being mapped to but rather the Ensembl ID that is connected to both the gene in the pathway and the gene in your dataset. Hence, for the node to color, your imported ID must map to the same Ensembl as the one the node in the pathway maps to. If you click on the node in the pathway you will see which gene it maps to in the backpage display on the right. Likewise, if you look at all nodes in your dataset and view all attributes in the datapanel you can see which Ensembl the gene of interest maps to. This may be resolved by using your probeset IDs as the primary ID when importing your datatset.

Best,
Nathan