Test Freebayes on sample files

Question

[Bella_p]

Hi,

I would like to test Freebayes variant calling that is part of my pipeline. Are there any available test files I can use to check that I get the correct results? Also for other variant callers?

[ATpoint]

Correct in terms of "correctly processed files" or correct true-positive variants?

[Bella_p]

What is the difference? If the files were correctly processed, doesn't it mean that I should get true-positive variants?

[WouterDeCoster]

If the files were correctly processed, doesn't it mean that I should get true-positive variants?

There is zero guarantees, and it is even extremely unlikely that all your variants will be true-positive.

[WouterDeCoster]

any available test files

You are looking for an example bam file?

I get the correct results

How will you check if the results are correct?

[Bella_p]

Yes, I'm looking for a bam file that I know in advance the variants it should find I guess. How would you test if a varaint caller you use is giving you a sensible results and not false-positive or false-negative variants?

[WouterDeCoster]

Search for Genome in a bottle (GIAB) data. For that sample data is freely available, together with a set of truth variants.

How would you test if a varaint caller you use is giving you a sensible results and not false-positive or false-negative variants?

Truth is, there will always be false positive and false negative variants. You just have to find a pipeline which minimizes those. But in general there is always a trade-off between sensitivity and specificity: would you rather miss a variant but have high precision, or are you okay with having some false positives but with a high recall?

[Bella_p]

Great, I'll try that. thanks a lot!