Huge thanks to Eva (shared first author), and to Wolfgang & Daniel, who supervised the project 🙏
🧵9/9

What became very clear throughout the project is how complex and unpredictable a, on the face of it, simple protocol can be. I think we mostly underestimate the complexity of molecular protocols.
Also: there’s room for discussion and research into what we consider “usable” reads (discussion).🧵8/9

We think our “funnel strategy” of maximizing usable reads could also be applied more broadly — as a QC metric and as a systematic way to optimize other bulk and single-cell RNA-seq protocols.🧵7/9

We combined these changes into prime-seq2. Across samples and libraries, it delivers +60% usable reads compared to prime-seq.
If you’d like to try it: we published a detailed, step-by-step protocol 🧪
👉 dx.doi.org/10.17504/pro...
🧵6/9

For prime-seq, the main loss points were:
• total sequencing reads
• index assignment
• intergenic reads
To improve this, we tested many alterations and ended up with:
✅ optimized DNA digest
✅ introduced PTO bonds into PCR primers
✅ redesigned ligation adapters & library PCR primers🧵5/9

This process can be nicely visualized as a funnel: at each step, reads are lost. The funnel makes it clear which stages are most responsible for read loss — and therefore where improvements could make the biggest impact.🧵4/9

We think what matters is the number of reads that can be used for analysis, "usable reads", while controlling for traditional QC metrics.
The usable reads number depends on:
- valid reads from the flow cell
- correct index
- filtering
- correct barcode
- exonic/intronic assignment
- unique UMI🧵3/9

Prime-seq has been used extensively in our lab. Over time, several imperfections became clear. During my Master’s thesis, I started on testing many potential improvements. To decide which alterations to keep, we needed a clear QC metric.🧵2/9