low mappability of scRNA data

Question

[annaA]

Hey everyone,

I am analyzing scRNA data from Smart2-seq . I did the mapping by using Hisat2 .Some of the cells has reaaly low mapping rate (8%) us expected. I am wondering do I need to exclude these cells before the following steps ( filtering with samotools, featureCounts) or they can be excluded later in Seurat? If I need to exclude them prio in further analysis I should use a specific tool ?

Thanks in advance, A.

Answer 1

[ATpoint]

I am not sure if Smart-seq is special but typically you perform alignment or quantification and make a count matrix. This is then inspected towards low-quality cells, e.g. by plotting the total number of detected genes, the total read or UMI count, the percentage of reads aligning to mitochondrial and ribosomal genes. This is covered in the Seurat (and basically any other) workflow. Low alignment percentage (if based on poor cell quality) should then be reflected in these metrics. I suggest you follow a guided tutorial if you are new to this type of analysis.

[annaA]

Hello, Thanks for your reply and the info.Smart2-Seq doesn't have UMIs, that's the only difference I have found some online tutorials in which it's not clear if I need to remove cells with low mappability prior to downstream analysis. But us you said and based on Seurat's tutorials probably I don't need to. Do you have any specific tutorial to recommend ?

A.

[ATpoint]

As said just follow the guided tutorials on the Seurat website and apply the filters they recommend. Alternatives are the workflows from the scater, scran packages at Bioconductor or the Bioconductor single-cell (OSCA) workflow.