Exposing the Machinery of the Resistome

aarondubrow writes "2011 Nobel Prize Winner, Bruce Beutler, is using the Ranger supercomputer at The University of Texas at Austin for an ambitious new project to discover all of the genes involved in the mammalian immune response – the so-called 'resistome.' Over several years, Beutler's lab will sequence the protein coding portions of genes in 8,000 mice to detect the impact of mutations on immunity. This means scanning, enriching and sequencing 500 billion base pairs every week. The project represents a 'Big Data' problem of the highest order."

Comments

Acquired data vs. archived data set

[fragMasterFlash]

The project represents a 'Big Data' problem of the highest order.

Before or after de-duplication of the data? Before, yes obviously but if that is still the case after de-duplication then gaining much knowledge from this experiment may prove to be a fools errand.

Re:Acquired data vs. archived data set

[Anonymous Coward]

I'm not sure the data is even that big. Given that a single base can be encoded in 2 bits, 500 billion base pairs could be stored in about 0.9 TB. The 8000 mice will have a lot of DNA in common with each other, so compression should be able to reduce storage requirements by several orders of magnitude..

Perhaps I am overlooking something, but this hardly seems like a "'Big Data' problem of the highest order". Could anyone with experience in the field of DNA sequencing please confirm or explain why I'm wrong?

Re:Acquired data vs. archived data set

[Anonymous Coward]

This isn't very much data. The analysis for an individual mouse exonerating can take less than 1/2 hour on a commodity CPU. Including identifying mutations and small insertion/deletion errors. So maybe 4k hours of computing. They should really be doing this with full genome resequencing, rather than exome capture- they'll miss lots of potentially interesting mutations.

My prediction

[Anonymous Coward]

Eleventy bazillion base pairs scanned... and still no Jebus in there. Just watch.

Holy fuck, sell that shit

[Anonymous Coward]

the software that solves the 'big data' problem of highest order, silly!

General problem

[Impy the Impiuos Imp]

Ok, quick question -- how do they determine what set of chemical markers in DNA constitute a "gene"? It seems like that could only be known by outcomes research by "running" the DNA to see what each little chunk produced.

Re:General problem

[Anonymous Coward]

The genetic code has comments. Actually, it has something like a boot record for each gene. The gene part is called an Open Reading Frame (ORF) and it is marked by stop codons. The gene is the part of the DNA that is to be transcribed by RNA and then sent to the ribosomes, which are 3D protein printers. There are little switches that turn this process on and off for the different genes. Some of the genes, such as for metabolism, run all the time, others are for special occasions.

Re:General problem

[Anonymous Coward]

The mouse genome is pretty well annotated in terms of where genes are. One can use gene prediction algorithms combined with sequencing of RNA to identify most or all of the genes in a species, and this has been done over the years for mice. In this case, they're doing whole-exome sequencing, where you enrich for known regions of the genome. That means they have a LOT less data to deal with (i.e. less aligning and SNP calling per mouse), which makes the experiment more tractable. On the down-side, that also means they'll miss a huge amount of important effects. But, this is probably meant to be a first pass, so that's ok.

Re:General problem

[mlush]

As others have pointed out it is possible to deduce where genes are by looking at the sequence however this is by no means straightforward DNA is spaghetti code of the very worst kind.

It is possible to "run the DNA to see what it produces", basically when a (DNA) gene is active copy's of its sequence are made in messenger RNA (mRNA, its like DNA but much less stable) the mRNA copy's are perhaps akin to compiled code as there is a fair amount of rearrangement that goes on before its '3D printed' in protein.

Now its possible to take the mash up cells or tissue extract the mRNA convert it back to DNA (stabilising it) then sequence the different DNA molecules to find out what genes are active in a given tissue.

Mapping the sequences back to the genes is not straight forward, genes can exist in multiple identical or near identical copy's making it hard to map back, some genes never express or only express under very specific circumstances or only in certain people

Anyone?

[Anonymous Coward]

Beutler? Beutler? Anyone? Beutler?

The what?

[bhcompy]

The Resisty? I mean, they had interesting technology, but the Irkens totally outclassed them.

Re:The what?

[Anonymous Coward]

Who cares about class.
They won because they had better snacks.

Re:We think that Mice are Reptiles

[Anonymous Coward]

... So, by process of elamination, ...

Was trying to think of a funny joke surrounding elamination, but the web search results were interesting (when you are bored at work) in their own right.

The only search result I got on elamination was this translation dictionary (http://en.bab.la/dictionary/english-spanish/elaminate):
  elaminate [bot.] (also: without lamina)

Lamina, in turn, comes back as type of spider (amongst other vague references and dead links):
http://en.wikipedia.org/wiki/List_of_Desidae_species

I've asked the mice what they thought about being reptiles and got nothing back except for a couple of squeaks. Looks like we will need to follow up with Tom Cruise in order to get a translation anyways.

- Toast

(OK, yes this post contributes nothing to the discussion, carry on!)

Ome My God

[gringer]

This was getting silly a few years ago with the metabolome. How many more omes (i.e. subsets of the total system that influences human biology) do we need to look at until we declare our human model complete? Is there going to be a 'humanome' that describes human-associated environmental factors? What about a 'radiatome' that describes the plethora of electromagnetic signals that enter our body over the course of a lifetime?

Re:Ome My God

[feepness]

Clearly we need to cull the number of "-omes".

I suggest a... thunderdome!

Two -omes enter! One -ome leaves!

Re:Ome My God

[Anonymous Coward]

I drew the line at Gnome.

Re:Ome My God

[Anonymous Coward]

Watch out for masterblasteromes.

Re:Ome My God

[Anonymous Coward]

There was the "proteonomics", the study of proteins, "bio-genomics", which was the study of the interaction of bacteria and human genes (original scientific thought was that the stomach was a sterile place due to the high levels of acid, then they discover that there are bacteria living there, and they communicate with the stomach lining using ghrelin. Scientists now have to study the interactions of bacteria and the human genome. There is now the situation where cells may have receptors to molecules that are not created by the host.

Going by all the different systems that a human body has (brain, autonomous/sensory/musculature control system, skin, muscle, digestive system, reproductive system, audio, visual, taste, smell, every one of those has unique cell configurations.

By the sound of things, we are descended from mice not monkeys.

Re:Ome My God

[Anonymous Coward]

"By the sound of things, we are descended from mice not monkeys."

With the exception of Christian Fundamentalists. The were created on the spot by some old bearded dude and have been going on our nerves since then.

Don't fuck with the immune system

[taumeson]

Is it just me, or is anybody else worried that the more we try to mimic the human body's immune system the more problems we are creating for ourselves vis a vis antibiotic resistance? We are so bad with how we use antibiotics that we are inevitably going to create bugs that are resistant to most forms of antibiotics. I surely don't want to create a class of bugs that are resistant to the VERY WAY THAT OUR IMMUNE SYSTEM WORKS.