Microsatellite Instability (MSI)

Question

[sruthi]

Hi,

Could anyone suggest me free tools for calculating MSI for tumor samples alone?

TIA

Answer 1

[Nicolas Rosewick]

You can use MSIsensor

• tool : https://github.com/ding-lab/msisensor
• article : https://academic.oup.com/bioinformatics/article-lookup/doi/10.1093/bioinformatics/btt755

Tumor only mode :

# first scan genome of interest 
msisensor scan -d reference.fa -o microsatellites.list
# Then use tumor bam file to detect MSI 
msisensor msi -d microsatellites.list -t tumor.bam -e bed.file -o output.tumor.prefix

[sruthi]

Hey Nicolas, Thanks for the reply. I have already used MSIsensor for 3 tumor samples. But I get 0 for "Number of sites with enough coverage" and "Number of somatic sites". How should I work further with MSIsensor?

[Nicolas Rosewick]

You should edit your question to add more informations regarding your data (which type : WGS, WES, targeted, etc... ). Without that I (and the community) would not be able to help you more.

[sruthi]

Yeah, sorry about that. I have used MSIsensor for 3 different tumor exome samples without matched normal and getting 0 as the value for "Number of sites with enough coverage" and "number of somatic sites (%)" for all the 3 samples. I need to know whether the value I'm getting is correct or am I missing out on something.

[Nicolas Rosewick]

Maybe check some sites in IGV to look if coverage is really low .. You could also reduce coverage thresold in msisensor mis with -c <int> coverage threshold for msi analysis, WXS: 20; WGS: 15, default=20

[sruthi]

I had set the coverage threshold to -c 20 but I still get the output as 0. Below is the code I've used, kindly let me know if I've done it the right way:

msisensor msi -d microsatellites.list -c 20 -i 1 -t tumor.bam -e bed.file -o output.tumor.prefix

And. I'll also check some sites with IGV.

[Nicolas Rosewick]

-c 20 is the default for WES. To decrease to min 10 of coverage try with -c 10

[sruthi]

I changed it to 10 and then increased the coverage as well. I'm still getting 0.

Answer 2

[sruthi]

Hi, Output of a tumor only case, is 2144 3491 15.01. I have a few queries regarding it.

1.From the sample output given on the github page, I understand that 15.01 is the % of "Number of somatic sites" and is the msi score as well. Is it msi-high or msi-low? 2. What is the cutoff score? Because it is mentioned that the cutoff values are calculated for various cancer types. 3.How do I get the MSI status?

TIA

[Nicolas Rosewick]

Figure 1 of the missensor paper seems explicit : https://academic.oup.com/bioinformatics/article/30/7/1015/236553 . As you have 15.01 % of somatic sites your sample should be MSI-H. The best would be to test on other samples ; and maybe on control samples (non-tumor) of the same tissue.

[sruthi]

Hey, Thanks for the article. I have read it. Let me explain from what I understood from the graph is, <10% would be msi-low and >10% is msi-high. Did I get it right?

[Nicolas Rosewick]

Yes indeed but should execute msisensor on all your samples and then check if you can see some stratification. Even better test it on control samples

[sruthi]

Okay will do it. Thank you so much.