DNA storage

[Brody]

A bioengineer and geneticist at Harvard’s Wyss Institute have
successfully stored 5.5 petabits of data — around 700 terabytes — in a
single gram of DNA, smashing the previous DNA data density record by a
thousand times.

http://www.extremetech.com/extreme/...rams-700-terabytes-of-data-into-a-single-gram

 

[Yousuf Khan]

A bioengineer and geneticist at Harvard’s Wyss Institute have
successfully stored 5.5 petabits of data — around 700 terabytes — in a
single gram of DNA, smashing the previous DNA data density record by a
thousand times.

http://www.extremetech.com/extreme/...rams-700-terabytes-of-data-into-a-single-gram

A gram is quite a large amount of DNA actually.

Yousuf Khan

 

[Tom Del Rosso]

A gram is quite a large amount of DNA actually.

Even a gram of complete cells would be about 1 billion cells. It would be 1
cubic centimeter since they are mostly water and that's the density of
water. A gram of DNA would be roughly the same size. A rather large and
fragile molecule. They say it's chemically stable but you have to read it
mechanically so it would be fragile.

Since it would be too fragile, you'd need a lot of strands and a separate
reading mechanism for each. The DNA itself would be an insignificant
fraction of the size and mass.

Now if these guys actually recorded on a single strand and they can read it
repeatedly and fast then I'd be impressed.

 

[Franc Zabkar]

Now if these guys actually recorded on a single strand and they can read it
repeatedly and fast then I'd be impressed.

"While it took years for the original Human Genome Project to analyze
a single human genome (some 3 billion DNA base pairs), modern lab
equipment with microfluidic chips can do it in hours."

"The work ... basically treats DNA as just another digital storage
device. Instead of binary data being encoded as magnetic regions on a
hard drive platter, strands of DNA that store 96 bits are synthesized,
with each of the bases (TGAC) representing a binary value (T and G =
1, A and C = 0)."

AISI, the data transfer rate is 3 billion DNA base pairs (= bits) in a
few hours.

If a "few" can be taken to mean 10, then the transfer rate would be
10KB/s. If 1 hour, then 100KB/s.

So sequencing 5.5 petabits at 100KB/s would require 240 years ???

- Franc Zabkar

 

[Tom Del Rosso]

"While it took years for the original Human Genome Project to analyze
a single human genome (some 3 billion DNA base pairs), modern lab
equipment with microfluidic chips can do it in hours."

"The work ... basically treats DNA as just another digital storage
device. Instead of binary data being encoded as magnetic regions on a
hard drive platter, strands of DNA that store 96 bits are synthesized,
with each of the bases (TGAC) representing a binary value (T and G =
1, A and C = 0)."

AISI, the data transfer rate is 3 billion DNA base pairs (= bits) in a
few hours.

If a "few" can be taken to mean 10, then the transfer rate would be
10KB/s. If 1 hour, then 100KB/s.

So sequencing 5.5 petabits at 100KB/s would require 240 years ???

I guess the genome project is refering to the speed of just one
lab-on-a-chip for "a few hours".

For the data storage claim, if they had 100 separate strands that could be
read simultaneously by 100 mechanisms on 1 chip, and if they had 80 chips,
then they could match SATA speed of 6 Gb/s (actually 6.4 Gb/s assuming the
high estimate of 100KB/s).

That would achieve good speed, but not the capacity they describe. A gram
of DNA would need a lot more than the area of 80 chips if spread out flat.

I don't know how DNA moves through those chips. I guess it circulates in a
canal.

 

[Yousuf Khan]

"The work ... basically treats DNA as just another digital storage
device. Instead of binary data being encoded as magnetic regions on a
hard drive platter, strands of DNA that store 96 bits are synthesized,
with each of the bases (TGAC) representing a binary value (T and G =
1, A and C = 0)."

Also, what a waste?!? Using DNA code to replicate binary code? DNA
should be at least quadrinary code because it's got four values. This
just wastes half the values.

Yousuf Khan

 

[Yousuf Khan]

If any nucleotide could pair with any other nucleotide, there would
be a hexadecimal system (4**2 values possible for each pair).

But only A can pair with T, and only C can pair with G to have a stable
DNA strand.
From Wikipedia on DNA replication
(http://en.wikipedia.org/wiki/DNA_replication):

Interesting, then DNA is really a binary code too?

Yousuf Khan

 

[Tom Del Rosso]

If any nucleotide could pair with any other nucleotide, there would
be a hexadecimal system (4**2 values possible for each pair).

But only A can pair with T, and only C can pair with G to have a
stable DNA strand.

I assume that means one strand has only A and C and the other strand only T
and G, so no reversals.

It's surprising that it stays that way because if a mutation ever did
reverse a pair then it would keep replicating.