Arachne linked-read aligner

The lariat developers made the case that linked-read data improved alignment, especially over highly repetitive regions. This was shown to work quite well in Aedes aegypti link and other works. Lariat relied on input FASTQ files with a rather peculiar variant: interleaved with 9 lines per record pair. Additionally, the Chromium 10X design did not preprocess the linked-read barcodes out of the sequence, hence the use of barcode "whitelists".

10X Genomics discontinued their linked-read technology in 2019 and Lariat was abandoned then as well. Since then, new linked-read methods emerged, namely haplotagging, stLFR, and TELLseq. These new techniques use different chemistries, but most importantly, all of them remove the linked-read barcode from the sequence and use conventional FASTQ formats. Lariat still has tons of value, so the goal was to update it for current technologies. To prevent platform lock-in and promote unified data standards, Arachne does not directly support any specific linked-read FASTQ format. Instead, it expects the 'standard' linked-read data format, which is a consistent future-proof format following the internationally recognized FASTQ and SAM specifications. Arachne provides a lossless converter that accepts haplotagging, stLFR, and TELLseq FASTQ data to faciliate these conversions. Our hope and intention is the ubiquitous adoption of this data format across all current and future linked-read chemistries.