search

xjtu-omics / msisensor-pro

Microsatellite Instability (MSI) detection using high-throughput sequencing data.
Other
98 stars 22 forks source link

MSISensor normalization issue. Please help! #42

Closed seifudd closed 1 week ago

seifudd commented 2 years ago

Hi,

We are running MSISensor version 0.60 for some clinical trials using whole exome sequencing (WES) data.

We are using deduplicated BAM files as input. All other parameters set to default.

We noticed that we had significant differences in coverage between tumor-normal (higher coverage in tumor) reads thus, we turned on the normalization parameter.

After normalization, we noticed that the MSI Percent scores were magnitudes higher compared to the non-normalized version across all samples.

Is there an explanation for this? Is MSISensor 0.60 weighting rather than normalizing. We were not expecting the scores to have this drastic increase.

Thank you very much for all your help and developing MSISensor. Appreciate it!

Thanks, Fayaz.

PengJia6 commented 2 years ago

We suggest you use the non-normalized version! because the normalized version is not validated on real data.