https://nf-co.re logo
Join Slack
Powered by
<nf-co.re/oncoanalyser/2.3.0/docs/usage#…>
# oncoanalyser
m
s
Hi @Michael Heskett, once you have selected you set of training samples, you should be able to complete the targeted panel resource creation workflow with just the BED file and a tailored driver gene panel
If you can provide some more details about which parts of the process didn’t work or where you became stuck, we can give you some advice!
m
thank you stephen, I am going to try it again from the start and will let you know more details on what failed. I was previously using core-sarek which requires you to make ASCAT specific files and that is a nightmare. I realized that you dont have to do that here so it should be easier
s
No worries, and please do let us know how you get on
m
Hi Stephen, im not exactly sure how to make the MSI custom panel file for a whole exome. is there a whole genome version that i can subset to my panel?
"target_region_msi_indels" specifically
it is also unclear what to use for target_region_ratios file. i can see that the TSO500 panel just uses some default values so ill copy those
@Stephen Watts not sure if you get a notification for this so i just wanted to tag you. sorry for the invonenience and thanks for your help
s
Hi @Michael Heskett, thanks for the ping, there isn’t a straightforward way to do this currently but functionality will be built-in to generate the MSI INDEL file automatically as part of the panel resource creation workflow in a future release
This file is optional however, and when not provided the analysis will still succeed, but just without an MSI estimate in PURPLE
Here is some further discussion that may (or may not) be helpful: https://nfcore.slack.com/archives/C04CQ1LK7U1/p1771668503268539
m
thanks. I made my own panel and it seems to work but I got an error with orange in that it did not deduce the correct cancer type, and it thinks the sample is tumor-only when it definitely wasn't. I wonder if its related or not
s
This appears to be expected behaviour if I understand correctly, you won’t get the tissue of origin prediction from CUPPA since it does not run in targeted mode. And the 
cancer (DOID 162)
string in top left is just the default cancer type oncoanalyser uses for cohort comparison
m
Thanks! The expression level comparison to other similar cancer samples also didn't run which now makes sense I guess since it doesn't infer the cancer type. In the US targeted panel or whole exome is very common and WGS is rare
but I can try to reproduce that on my own using nfcore-rnaseq and nfcore-sarek
m
Hi stephen just folowing up on targeted mode: 
ORANGE 5.0.1 crashes with a null pointer because its PDF front page expects a CHORD result, which targeted mode doesn't produce
Tumor / normal targeted mode support would be awesome. its what thousands upon thousands of patients are doing right now
because consumer genomics companies are doing dna exome / rna seq paired
in fact at my research org, ive essentially had to rebuild and entire clinical pipeline just because oncoanalyser doesnt handle this well. I would much prefer to use your code since its well established
s
Thanks for the report, we'll address this in the 3.0.1 or 3.1.0 release, and I'll add a test case on my side to ensure that this functionality continues to work through releases
m
Thanks stephen!
I am in USA so time difference is bad but i would like to become an oncoanalyser contributor if possible. i can help. PhD and good with python (learned 10 years before AI coding)
2 Views
Open in Slack
PreviousNext
Powered by