Recently, I was asked a question by a customer: How can I find transcripts specific to prostate cancer? The added difficulty to this question was that the answer needed to assure that the transcript's expression was not tissue-specific and was confined to just Prostate Cancer.
Using Omicsoft's OncoLand and Land analytics, we were able to accomplish this in a few minutes.
- Group samples across all Lands, including GTEx (normal samples) into Prostate cancer vs All. Any sample from the PRAD grouping was labeled Prostate Cancer, and all other samples were labeled "All".
- Use Sample Grouping -> Expression Land Analytics module to find alternatively spliced transcripts across these two groups.
- As a further refining criteria, run the analytical module again, this time just comparing the PRAD Tumor samples to the PRAD control normal samples.
- Merge the two results in Array Studio, then filter for a set of transcripts that both go in the same direction (i.e PRAD>Control and Prostate Cancer>All) with an mean fold change of >2.
This gave me a list of transcripts that can be considered alternatively spliced, specific to Prostate Cancer, including transcript uc003htj.4, which I show here.
Interestingly, expression of other transcripts for PDLIM5 were specific to other tissues, including Heart and Muscle-specific transcripts.