Pipeline for inferring tissues from liquid biopsy data

[SuzuBell]

In many cases, solid tissue biopsies cannot be performed in humans. If a liquid biopsy is performed on humans, we can assess microRNA, RNA, and DNA in the blood. These microRNA, RNA, and DNA presumably came from solid tissues. I have read that it can be inferred which tissues these microRNA, RNA, and DNA likely came from.

If that is true, what is the general pipeline for such an analysis (at a high level, i.e. for a non expert)? Are there any resources for a newbie interested in this type of research to better understand this pipeline?

[Michael.Ante]

For solid tissue analysis, you have the assumption, that your observed xNA (miroRNA, RNA, DNA, etc.) is derived from one origin showing a certain 'fingerprint'. Meaning that your observation is depending on the tissue type. In such a case, you can use e.g. biogps.org datasets to perform a GSEA to obtain/rank the tissues which fit best to your data set.

In your blood or other liquid sample, you have a mixture of xNAs produced in various tissues. The "simple" dependency assumption is not working.
I fear it's not as easy to trace back from which tissue a certain amount a feature is produced. Especially, since you cannot deduce directly from an observed tissue-related expression/feature that this tissue is able to secrete the xNA into the blood.

There might be certain microRNAs / RNAs which are only produced in certain tissue(s) exclusively (for instance MUP4 in mouse lacrimal glands).

You might start with a simple likelihood model for each feature from which tissue it might came from.

Best,

Michael

[SuzuBell]

Thank you for your response. I can certainly write some scripts that use likelihood models to estimate which tissue each feature may be derived from.

As a confirmation, it seems there are no software or published pipelines already available for inferring which tissues xNA from blood samples are likely derived from?