# De Novo Calling Queries

**URL:** <https://discuss.hail.is/t/de-novo-calling-queries/2614>\
**Category:** Hail Query & hailctl\
**Created:** [April 15, 2022, 4:05pm UTC](https://discuss.hail.is/t/de-novo-calling-queries/2614 "2022-04-15T16:05:15Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![genome](https://avatars.discourse-cdn.com/v4/letter/g/9f8e36/32.png) [@genome](https://discuss.hail.is/u/genome)\
**Post date:** [April 15, 2022, 4:05pm UTC](https://discuss.hail.is/t/de-novo-calling-queries/2614/1 "2022-04-15T16:05:15Z")

</div>

Hi all! I am trying to use the` hl.de_novo` function to call de novo variants. I am quite new to this, so can I get some insights if this function will suffice for making “good” calls, or if there is additional filtering that I should be doing prior to using the function?

Thank you very much !!

---

<div class="post-metadata">

**Author:** ![tpoterba](https://yyz2.discourse-cdn.com/flex036/user_avatar/discuss.hail.is/tpoterba/32/61_2.png) [@tpoterba](https://discuss.hail.is/u/tpoterba)\
**Post date:** [April 19, 2022, 2:52pm UTC](https://discuss.hail.is/t/de-novo-calling-queries/2614/2 "2022-04-19T14:52:21Z")

</div>

This function is supposed to be a general purpose de novo calling method, which uses the quality metrics and frequencies to build a statistical model of true positive probability. You shouldn’t need to filter ahead of time, though the method does not account for e.g. poor mapping quality of variants (you should filter to PASS ahead of time).

---

<div class="post-metadata">

**Author:** ![genome](https://avatars.discourse-cdn.com/v4/letter/g/9f8e36/32.png) [@genome](https://discuss.hail.is/u/genome)\
**Post date:** [April 20, 2022, 7:24pm UTC](https://discuss.hail.is/t/de-novo-calling-queries/2614/3 "2022-04-20T19:24:09Z")

</div>

Thank you Tim! To clarify, will filtering to PASS before the test affect the statistical model (compared to not filtering)?

---

<div class="post-metadata">

**Author:** ![tpoterba](https://yyz2.discourse-cdn.com/flex036/user_avatar/discuss.hail.is/tpoterba/32/61_2.png) [@tpoterba](https://discuss.hail.is/u/tpoterba)\
**Post date:** [April 20, 2022, 7:50pm UTC](https://discuss.hail.is/t/de-novo-calling-queries/2614/4 "2022-04-20T19:50:03Z")

</div>

No. This model runs per-variant independently, so filtering on variants beforehand will remove results from the final output, but the statistical model is the same.