5 ms·
With the Novaseq X the cost is approximately $200 a genome. One the current challenges is storing the hundreds of gigabytes of data per genome.
by gww 3y ago
With the Novaseq X the cost is approximately $200 a genome. One the current challenges is storing the hundreds of gigabytes of data per genome.
- touisteur 3y ago1TB SD cards are a thing for 150USD now? Maybe two for physical redundancy and some Reed-Solomon stuff on each card?
- gww 3y agoI wonder if that would be cheaper in the long term than the typical NFS plus off site tape backups and/or cloud backup used by the places I have worked. I suspect it would become a logistical nightmare to store and keep track of thousands of SD cards.
- ojosilva 3y agoI found the issue about storage quite interesting, went digging and I found this article: https://medium.com/precision-medicine/how-big-is-the-human-genome-e90caa3409b0 https://medium.com/precision-medicine/how-big-is-the-human-g... Although the article does not mention it anywhere, I thought such repetitive byte sequences could be compressed quite well with simple file compression. Went digging again. I was wrong, compression of FASTQ genome sequence files IS a thing: https://en.wikipedia.org/wiki/FASTQ_format#Compression https://en.wikipedia.org/wiki/FASTQ_format#Compression
- jghn 3y agoIt goes far beyond that. By the time any real analysis is done on these data, they're extremely compressed via other file formats that extract the important signal. Alignment data is stored in BAM/CRAM [1], [2] formats, and variant data in VCF [3] [1] https://en.wikipedia.org/wiki/Binary_Alignment_Map https://en.wikipedia.org/wiki/Binary_Alignment_Map [2] https://en.wikipedia.org/wiki/CRAM_(file_format) https://en.wikipedia.org/wiki/CRAM_(file_format) [3] https://en.wikipedia.org/wiki/Variant_Call_Format https://en.wikipedia.org/wiki/Variant_Call_Format