Topic: ASIC miner for genomic sequence alignment (Read 819 times)
Hiya colleague! I'd like to know if it's possible too. Gotta do some research since I'm waiting for my ASIC to arrive soon.
I love the idea. I can't claim to know about the intimate details of how an ASIC works, but from a lay person's understanding; an ASIC solution would fit the requirements well in your cause, however, an ASIC designed for mining wouldn't likely work as the instruction set is built into the chip. I don't believe it would be reprogrammable to work for genome sequencing.
Hello,
I am a scientist and some of the work I do involves next-generation sequencing data.
The way this data is generated is that short pieces of sequence 'reads' of about 100 bases (letters) are generated in a massively parallel way and then aligned (matched) to a reference genome. The human genome, for example, is about 3 billion bases long. This is, of course, a computationally intensive process and is usually carried out on a cluster or in the cloud.
What I was wondering, is whether this process could be carreid out using an ASIC miner (probably a scrypt ASIC). The tools usually used for alignments include BWA (http://bio-bwa.sourceforge.net/bwa.shtml) and Stampy (http://www.well.ox.ac.uk/project-stampy) in case anyone is interested in answering this question. By way of incentive, this is a multi-billion$ industry...
*waits for sequence to align for the next few weeks*
I am a scientist and some of the work I do involves next-generation sequencing data.
The way this data is generated is that short pieces of sequence 'reads' of about 100 bases (letters) are generated in a massively parallel way and then aligned (matched) to a reference genome. The human genome, for example, is about 3 billion bases long. This is, of course, a computationally intensive process and is usually carried out on a cluster or in the cloud.
What I was wondering, is whether this process could be carreid out using an ASIC miner (probably a scrypt ASIC). The tools usually used for alignments include BWA (http://bio-bwa.sourceforge.net/bwa.shtml) and Stampy (http://www.well.ox.ac.uk/project-stampy) in case anyone is interested in answering this question. By way of incentive, this is a multi-billion$ industry...
*waits for sequence to align for the next few weeks*
An ASIC miner is made out of lots and lots of specialized integrated circuits that have been customized to do one thing only: hashing at very high speeds. Since this type of MCU does not waste any cycles on anything else it is very efficient compared to, say, a personal computer processor which does a lot of other things at the same time.
Unless you can break down your sequencing algorithm to solving lots of SHA256 hashes, you can't use a mining ASIC for that.
You should look into specialized ICs if already designed, if not (likely) you can try to contact an electronics engineer to help you design some FPGA-based boards for your specific algorithms. A FPGA based solution will be much faster than what your current options are (mainly the hyped and overpriced cloud) and if you get some VC you can design your own ASICs and break the world record for processing genome data.
Keep in mind that most popular VC companies will not fund you because they're likely in cahoots with certain companies and other groups of interest that would be very keen (multi-billion dollar worth of keen) to keep ripping off their pharma clients and specialized in blowing away R&D budgets through the Seychelles.
So your best bet is to hit it off with a small group of literate private investors (not some random hillbilly who just found oil in their backyard) and invest significant effort in your qualifying process.
Contact this person if you're really serious with your research: https://twitter.com/rikisabatini
Hello,
I am a scientist and some of the work I do involves next-generation sequencing data.
The way this data is generated is that short pieces of sequence 'reads' of about 100 bases (letters) are generated in a massively parallel way and then aligned (matched) to a reference genome. The human genome, for example, is about 3 billion bases long. This is, of course, a computationally intensive process and is usually carried out on a cluster or in the cloud.
What I was wondering, is whether this process could be carreid out using an ASIC miner (probably a scrypt ASIC). The tools usually used for alignments include BWA (http://bio-bwa.sourceforge.net/bwa.shtml) and Stampy (http://www.well.ox.ac.uk/project-stampy) in case anyone is interested in answering this question. By way of incentive, this is a multi-billion$ industry...
*waits for sequence to align for the next few weeks*
I am a scientist and some of the work I do involves next-generation sequencing data.
The way this data is generated is that short pieces of sequence 'reads' of about 100 bases (letters) are generated in a massively parallel way and then aligned (matched) to a reference genome. The human genome, for example, is about 3 billion bases long. This is, of course, a computationally intensive process and is usually carried out on a cluster or in the cloud.
What I was wondering, is whether this process could be carreid out using an ASIC miner (probably a scrypt ASIC). The tools usually used for alignments include BWA (http://bio-bwa.sourceforge.net/bwa.shtml) and Stampy (http://www.well.ox.ac.uk/project-stampy) in case anyone is interested in answering this question. By way of incentive, this is a multi-billion$ industry...
*waits for sequence to align for the next few weeks*
Jump to: