# SMART-seq and MARS-seq pipelines

**URL:** <https://discourse.scverse.org/t/smart-seq-and-mars-seq-pipelines/3681>\
**Category:** scvi-tools\
**Tags:** integration\
**Created:** [June 10, 2025, 3:00pm UTC](https://discourse.scverse.org/t/smart-seq-and-mars-seq-pipelines/3681 "2025-06-10T15:00:31Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Roy](https://avatars.discourse-cdn.com/v4/letter/r/4da419/32.png) [@Roy](https://discourse.scverse.org/u/Roy)\
**Post date:** [June 10, 2025, 3:00pm UTC](https://discourse.scverse.org/t/smart-seq-and-mars-seq-pipelines/3681/1 "2025-06-10T15:00:31Z")

</div>

Hi all,

I’ve been working on a Python-based single-cell pipeline for cross-study integration built around the scverse ecosystem. So far, it works well for 10x Genomics data, but I’m now looking to expand support to other technologies like MARS-seq and SMART-seq.

I am facing some challenges with MARS-seq: using similar QC pipeline to what I use for 10x data (even with adjusted thresholds) seems to be too stringent, I suspect this is due to lower UMI complexity and transcript capture efficiency in MARS-seq compared to droplet-based methods.  
SMART pseudo-bulk nature also poses some challenges.

Has anyone here successfully used scverse-based tools on MARS-seq or SMART-seq data? Interested in ways to adapt QC, ambient removal (SCAR particularly) to work with these.

Any insights or examples would be greatly appreciated!  
Roy

---

<div class="post-metadata">

**Author:** ![cane11](https://yyz1.discourse-cdn.com/flex035/user_avatar/discourse.scverse.org/cane11/32/241_2.png) [@cane11](https://discourse.scverse.org/u/cane11)\
**Post date:** [June 19, 2025, 5:56pm UTC](https://discourse.scverse.org/t/smart-seq-and-mars-seq-pipelines/3681/2 "2025-06-19T17:56:00Z")

</div>

SMART-seq, yes definitely in e.g. Tabula sapiens. We also provide tutorials for Smart-Seq2 in scVI-tools and I also used MARS. However, I never combined it with SCAR and obviously changed the QC thresholds for these datasets. Could you provide a clearer error m?