# Iterative use of nextstrain or parameters tuning

**URL:** <https://nextstrain.discourse.group/t/iterative-use-of-nextstrain-or-parameters-tuning/631>\
**Category:** Help and Getting Started\
**Created:** [July 23, 2021, 7:41pm UTC](https://nextstrain.discourse.group/t/iterative-use-of-nextstrain-or-parameters-tuning/631 "2021-07-23T19:41:51Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![qwerty123](https://avatars.discourse-cdn.com/v4/letter/q/54ee81/32.png) [@qwerty123](https://nextstrain.discourse.group/u/qwerty123)\
**Post date:** [July 23, 2021, 7:41pm UTC](https://nextstrain.discourse.group/t/iterative-use-of-nextstrain-or-parameters-tuning/631/1 "2021-07-23T19:41:51Z")

</div>

Dear all,

I faced with the following problem:  
Building the tree takes too long due to the large amount of data I have. Now it takes a week to calculate it. And with each next week, the amount of data increases. (Now I have already 20k of subsampled samples, but soon it will be about 40-50k)

(I am still using the default parameters for Nextstrain. In other words, I haven’t changed anything in main\_workflow.smk)

Is it possible to somehow optimise (or tune) some of the parameters in one of the .smk files to speed up the calculations?

Or maybe I can build a new tree not from scratch, but based on the previous one.  
For example, I have my already built tree. I am getting a new data. Can I build a new tree with new data starting from the tree I already have?

---

<div class="post-metadata">

**Author:** ![rneher](https://yyz1.discourse-cdn.com/flex031/user_avatar/nextstrain.discourse.group/rneher/32/20_2.png) [@rneher](https://nextstrain.discourse.group/u/rneher)\
**Post date:** [July 26, 2021, 8:27pm UTC](https://nextstrain.discourse.group/t/iterative-use-of-nextstrain-or-parameters-tuning/631/2 "2021-07-26T20:27:20Z")

</div>

thanks for reaching out.

Unfortunately, neither the processing pipeline `augur`/`treetime` nor our visualization is well suited to handle more than 10-20k sequences. There are a few ways in which things could be sped up (tree building parameters, skipping confidence calculation for the timetree (or the time tree calculation altogether)). But to give specific advice, we would need to know more about your use case.