We did mapping paired-end (PE) 3' 10X (scRNA-seq data) reads with a ref genome using STARsolo, but mapping result shows a significant amount of unmapped reads. We want to extract the unmapped reads, assemble them and perform a blast search to specify the other species if any.

I have 2 questions-

• Is there any specfic code or algorithm to extract the unmapped reads of PE 3' 10X data mapping since its Read1 i.e. R1 has umi+bar code and only R2 has sequence? OR should we use samtools to extract them?

• Same question is for assembly. i.e. Should we use a de novo assmbler
  for assembling the unmapped reads of above data? or is there any
  specific tool for that since it's 10X data?

Thanks for your time and help.