Does disabling error correction lead to a loss of reads?

[wlandesman (Bill Landesman)]

Does this approach (i.e. not running error correction due to having 10nt barcodes) result in a loss of data if an error is present in a bar code?

[thermokarst]

Hey there @wlandesman!

In short, yes. Demultiplexing using an exact match approach. Check out this post I recently wrote on the topic.

emp demultiplexing: No sequences were mapped to samples

Hi @Yavnika! The safest way is to check with your sequencing center! Check with your sequencing center to be sure. Error correction helps you recover more sequences per sample by correcting any read errors in the barcode. With correction disabled you will lose any sequences where the barcode has an error in it (an A called as a C, for example). If you don't have many errors in your barcodes, then error correction might not be necessary. If you're comfortable with the percentage of your reads…

I hope that helps!

[wlandesman (Bill Landesman)]

Thanks for the reply! Are there options out there for working with 10bp barcodes?

[thermokarst]

You can use the emp-* methods in q2-demux or the demux-* methods in q2-cutadapt to demultiplex barcodes of any length, but for Golay barcode error correction, we currently only support EMP barcodes, which are 12nts long.

[wlandesman (Bill Landesman)]

OK that's what I thought. Funny you should mention cutadapt - just tried that for the first time yesterday for cutting off my adapters. Thanks again!