What the use of MEDIAN_CV_COVERAGE from picard RNA metrics

Question

[pengchy]

Hi all,

I have calculated the MEDIAN_CV_COVERAGE, however, what the meaning of this measurement? I have checked the related discussion at seqanswers and asked the question there. In brief, if the MEDIAN_CV_COVERAGE as the defined on wiki, the calculated value should be larger than the output of picard. In my data sets, this value just range between 0.2 and 1.7. To my understand, this value will definitely larger than 0 for normal samples. Because the coverage for high expressed genes should be very high and the variation for the top 1000 highly expressed genes will be large. Am I right?

Thank you!

Best,
p

Answer 1

[Sean Davis]

The MEDIAN_CV_COVERAGE represents a measurement of the variability of coverage over single genes. Ideally, there would be no variation over the length of the genes, but in reality, there is always some. Extreme outlier samples may represent technical differences in those samples that lead to higher variability. In practice, take a look in a browser (IGV, etc.) at the extreme samples on the low and high end to get a sense of what the value represents.

[pengchy]

Hi Sean Davis, Thank you for your answer. I have misunderstood this for variation of expression level across genes.

I wonder how the coverage was calculated, because the value of MEDIAN_CV_COVERAGE ranges from 0.2 to 1.7 in my samples. I guess the coverage of one gene is calculated as the coverage CV across all positions.

[Sean Davis]

You are correct.