Comparing Pathways Between Two Different Cancer Cell Lines

Question

[Woa]

Can somebody let me know how to compare differences in pathways between two cancer cell lines? The two cell lines have one or two key proteins present/absent in them. We have estimates of the relative protein expression levels from proteomics data. How the analysis would be different from comparing cancer vs. normal cells by enrichment/pathway analysis of over and underexpressed proteins? Any pointers to relevant article, review paper, software, statistical method is most welcome.

Thanks in advance

WoA

Answer 1

[Khader Shameer]

I have been working on similar analysis for pairs of diseases.

I would do it this way:

1. Take cell line 1 (for example consider "AA" as cell line 1)
2. Extract all genes that are present
3. Input list of genes here (Reactome Pathway Analysis)

Click on Overepresentation analysis. Finds the Reactome pathways in which IDs in your list are strongly enriched - can help to understand the biological context of your data. You will get a list of pathways (or events) with following fields

• False discovery rate
  
• Un-adjusted probability of seeing N or more genes in this Event by chance
  
• Number of genes in your query which map to this Event
  
• Total number of genes involved in this Event
• Identifier of this Event
  
• Name of this Event
  
• Submitted identifiers mapping to this Event

Select pathways that are statistically significant (P <= 0.05)

Repeat the same for cell line 2 (AT) and get the statistically significant pathways.

Create a Venn-diagram and see the pathways shared between the two cell lines.

Here is an example Venn-diagram that shows overlap between pathways

Now you can apply set-theory to make sense of your data. [?] For example in the figure: AA ∩ AT = Common between two cell lines[?]
Relative complement of AA = Pathways specific to AT (Purple, n=4)[?] Relative complement of AT = Pathways specific to AA (Yellow, n=216)[?]

NB: You may need additional statistical analysis to show that the pathways shared between cell lines or private pathways are statistically significant and not random.

Caveats:

• Reactome consider pathways as 'events'
  

• Reactome may not identify all of your input gene names. Use an ID mapping service or Reactome mapping service to map the identifiers.

• Reactome provide un-adjusted p-values

• You may also do an under-representation analysis to see what all putative pathways could be perturbed due to absence of several genes in the cell lines.

• You can do similar type of analysis using other tools like WikiPathways.

Also see similar discussions on pathway analysis or pathway enrichment in BioStar

[Gjain]

i can`t see your venn diagram

[Khader Shameer]

Thanks @Gjain, I fixed the issues with the figure.

[Gjain]

Thanks, very interesting.

[Woa]

Many thanks Khader

Answer 2

[Gjain]

you can also try BioCyc