Question: Reactome pathways in ReactomeDB/graphite
gravatar for i.sudbery
2.9 years ago by
European Union
i.sudbery10 wrote:

ReactomePA pathway plots generated using viewPathway seem to be missing components of the pathway.

For example if you view the human "HDL-mediated lipid transport" using viewPathway, the resulting network has 14 nodes. However, if you get any of the pathway mapping files from the reactome website, the "HDL-meidated lipid transport" pathway in humans has 19 members. So for example the genes "ALB", "HDLBP" and "PLTP" are listed as being members in the NCBI2pathway mapping from the reactome website, and on the website webapp, but don't appear on pathView.

 I've tracked down that ReactomePA pulls its Reactome data from graphite, and that when you pull the "HDL-mediated lipid transport" out of graphite, it does indeed only contain these 14 genes (actaully they are UniProt IDs in graphite). The data retrieved is marked as 12-10-2016, so it seems unlikely that the data is stale.

Any idea where this data came from, and why its different to that on the website.



graphite reactome reactomepa • 705 views
ADD COMMENTlink modified 2.9 years ago by gabriele.sales100 • written 2.9 years ago by i.sudbery10

Thank you for bringing this up.
I also checked the reactome.db package and it also does not list the same gene ids as in the NCBI2Reactome.txt file.
Interestingly, I have worked with people from Reactome to get the query correct. (could be that the query does something similar as mentioned below)
I will see if I can depend on those files for a future release. That should decrease inconsistencies.

ADD REPLYlink modified 2.8 years ago • written 2.8 years ago by willem.ligtenberg150
Answer: Reactome pathways in ReactomeDB/graphite
gravatar for gabriele.sales
2.9 years ago by
European Union
gabriele.sales100 wrote:

Dear User,

graphite developers here. We've taken a look at the "HDL-mediated lipid transport" pathway, and in particular at the ALB protein (P02768).

In the original Reactome pathway such protein takes part to the following biochemical reaction:

ALB + 1-acyl LPC -> Complex(ALB + 1-acyl LPC)

which encodes a physical interaction between the protein itself and a small molecule to form a new complex. Such reaction is isolated: none of ALB, 1-acyl LPC or their complex appears in any other process.

The main focus of graphite is the generation of protein networks that can be used to implement topological pathway analyses. In order to obtain them, the software reads Rectome BioPAX files and performs a number of transformations including:

  1. Removal of small molecules, possibly preserving signals going through them (not needed here). This step is required because we want the final network to contain proteins only.
  2. Removal of self loops and isolated nodes, because they tend to be incompatible with topological analyses (for instance, clipper or SPIA).

Because of (1), the reaction above becomes:

ALB -> Complex(ALB)

which is trivially


This self-loop is then pruned by (2), and this explains why you can no longer find such protein in the transformed network.

You can read more about the conversion process implemented by graphite here:


  Enrica & Gabriele

ADD COMMENTlink written 2.9 years ago by gabriele.sales100

thank you! very clear.

ADD REPLYlink written 2.9 years ago by Guangchuang Yu1.1k

Thanks, that makes perfect sense! It would then seem that using graphite as input to ReactomePA is possibly inappropriate as several of the analyses relie in part on enrichment analysis or perhaps it is a bit weird that such a reaction is listed as part of the "pathway", when really it just means that it is connected to the process.

ADD REPLYlink written 2.8 years ago by i.sudbery10
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 174 users visited in the last hour