I am working on a project to detect genetic mutations in lung cancer using machine learning algorithms and I am new to genetics and medical sciences. Can I use RNA as input to the algorithm for diagnosis? In medical diagnosis, do they check RNA first and then identify genetic mutations?
This is a nuanced question with a nuanced answer. By RNA I presume RNA fusions. If that is the case, it helps to consult popular guidelines such as ESMO, NCCN and AMP/ASCO for testing recommendations. NCCN guidelines previously stated that:
"Guidelines from the National Comprehensive Cancer Network (NCCN) recommend that, to maximize the detection of fusion events in patients with NSCLC, RNA-based sequencing should be used if no driver oncogenes can be identified in broad panel testing (NCCN Guidelines, version 6.2022)."
That being said, you can do both at the same time resources permitting:
"e.g., for a NSCLC you would want the test to minimally cover mutations (EGFR, KRAS, MET, BRAF, and ERBB2) AND detects fusions (ALK, RET, ROS1, NTRK1, NTRK2, and NTRK3" (from - https://education.nccn.org/sites/default/files/course/2023-05/00_Morrissette_NCCNbio23_v1.pdf)
Usually, for lung cancer you go for DNA first because it is more likely the patient has variant(s) in EGFR, KRAS, BRAF etc, or CNVs in these genes. Gene fusions in ALK, RET, NTRKs are less prevalent but therapeutically highly relevant so if the DNA test is negative for actionable variants, it is worthwhile to check for RNA fusions. Labs can also do both DNA + RNA at the same time if resources permit. There are multimodels assays that help do both nowadays and this way of analysis is gaining traction as it offers a comprehensive solution without the need for retesting and resampling.
Thanks for your helpful answer.
Therefore, we should not use RNA data to detect EGFR mutation in NSCLC. Is it correct? Can Justin Kiessling
Are you trying to predict DNA variants given RNA fusions? That could be a useful study, even if it yields no insights. Then we'd know that RNA fusions cannot predict DNA variants (we won't know that until we see the data). Even a study with negative results is informative, we shouldn't bias to only positive findings. Would even be interesting to see if RNA fusion positive samples exclude particular DNA genes or DNA variants. For example, in EML4-ALK fusion positive patients do we never see EGFR positive samples? What about the opposite, do some DNA variant positive sample profiles exclude RNA fusion?
Thank you for your comprehensive answer.
Yes. I am trying to predict DNA variants given RNA fusions. But, I want to know: Does EGFR mutation have a direct effect on the patients' RNA or not? Has a study been done in this field?
Can Justin Kiessling
That's an interesting question that I don't have an answer for. Some things to think about, a lot of the NGS studies do bulk sequencing, that is they sequence a tumor sample and give you a global reading - you don't know if a particular DNA variant is in cis configuration with a gene fusion in the same cell. The DNA variant + RNA fusion could be in the same cell, or perhaps in completely different cells. You could however discern this with single cell sequencing. Remember, cancer is a heterogenous disease, there are different clones and subclones which share some common mutations as well as mutations that are private to a subclone. Have a look at single cell studies and see if you see fusions and oncogenic variants together in the same cell. RNA fusions can sometimes be a major driver that you don't need many onocogenic variants in the tumor, but as with cancer this doesn't always pan out due to its heterogenous nature.
- Badges
- Science method
- Similar topics
- Biological Science
- Histology
- Staining