Free Program Predicts How Troublesome a Genetic Mutation Is

smoothjazz writes "Researchers describe a new, freely available Web-based program called Spliceman for predicting whether genetic mutations are likely to disrupt the splicing of messenger RNA, potentially leading to disease. From the article: 'Spliceman makes its predictions about mutations by calculating that distance. It has successfully predicted the known effect of many mutations. The software has genomic information about 11 species: humans, chimpanzees, rhesus monkeys, mice, rats, dogs, cats, chickens, guinea pigs, frogs, and zebra fish.'"

Comments

Inbreeding

[j. andrew rogers]

This app is part of the sex education class in Arkansas!

Re:Inbreeding

[Samantha Wright]

What sex education class in Arkansas?

Re:Inbreeding

[K. S. Kyosuke]

Arkansas...that's the state with that weird pronunciation of its name, "arcane sex" or something like that?

static code analysis tool

[vlm]

Rephrased, its a static code analysis tool for mRNA

http://en.wikipedia.org/wiki/List_of_tools_for_static_code_analysis

Now taking bets on 10:1 odds that after posting this, some joker adds spliceman to the wiki page. Frankly it probably does belong there.

Yes, I am one of "those guys" who trys to make my perl scripts Perl::Critic compliant, in addition to running them thru perltidy. The closest thing I know of for Ruby is "laser" and I don't know a good reformatting tool for Ruby.

Re:static code analysis tool

[Samantha Wright]

That's the idea. Of course, mRNA splicing sites are way harder to find than blocks of code (the sequence is described probabilistically, in essence), and it's subtly different between species. Fortunately it's not even remotely Turing-complete; it's closer to a context-free grammar, on the level of parenthesis matching. Which is evaluated in a probabilistic order. With Turing-complete compiler macros. (It's true: all living organisms are secretly written in PL/I.)

Re:static code analysis tool

[nani popoki]

That explains why DNA has so much bloatware and runs only on ancient hardware.

So let's make this a Game!

[ackthpt]

Call it Mutant Madness, try to find the most troublesome mutant genes, which are indifferent or compatible with other mutations and award skills, points, ...

Mmmm. Sounds too much like my massive collection of Crazy Bones. Nemminds.

Re:So let's make this a Game!

[Samantha Wright]

Humans aren't very good at the problem of RNA splice site finding, unfortunately. Like Reversi/Othello, it involves detecting a large number of subtle changes in closely-related states that have dramatic downstream effects. The game would consist primarily of looking at a bunch of roughly hill-shaped graphs and trying to determine which ones cross certain thresholds or most closely resemble certain other curves. Computationally those are pretty manageable tasks.

Re:Gamemaker!

[ackthpt]

Switch to Crowdsourcing to fix these problems!

Amateur scientists, amateur scientists! I can't even stand it anymore! These scientists don't want to use Crowdsourcing because then everything would change!

Scientists are unreliable; they're always changing their minds!

Shhh. Don't tell them it's a g-a-m-e. More fun this way. Also, if patents are involved and someone's profiting from the crowd's work, then it's helping Wall Street ;-)

Paging Professor Xavier

[sehlat]

This sounds like a great tool for predicting the abilities of new recruits.

Re:It's begun

[0123456]

Back in the real world, the 'bio age' will be another incentive to get the hell off this planet before someone wipes us out with an engineered disease. I'm sure the 'humans are a cancer' Greenists are just salivating at the thought.

Re:ha!

[Anonymous Coward]

So can this explain why my parents are so stupid?

I see that at least these genes were passed faithfully.

Ruined by stupid PR

[pesho]

Free Program Predicts How Troublesome a Genetic Mutation Is

No it doesn't.

This is yet another case of how stupid PR does damage to otherwise good science. The original paper (not the PR release) describes a statistical model. The model tells you how likely it is for a mutation to affect splicing. According to ROC curve on figure 4 it isn't nearly as selective or sensitive as you need it to be for any clinical application. It is however a great research tool. The PR article makes another false assumption, by stating that mutations affecting the splicing of an exon invariably cause 'trouble'. In fact it is well established that there is a significant variability in exon usage in perfectly healthy humans (citations below).

1. Kwan, Tony, David Benovoy, Christel Dias, Scott Gurd, David Serre, Harry Zuzan, Tyson A. Clark, et al. “Heritability of Alternative Splicing in the Human Genome.” Genome Res. 17, no. 8 (August 1, 2007): 1210–1218.
2. Zhang, Wei, Shiwei Duan, Emily O. Kistner, Wasim K. Bleibel, R. Stephanie Huang, Tyson A. Clark, Tina X. Chen, et al. “Evaluation of Genetic Variation Contributing to Differences in Gene Expression Between Populations.” The American Journal of Human Genetics 82, no. 3 (March 3, 2008): 631–640.

Re:Ruined by stupid PR

[zebslash]

As a geneticist, I completely agree. Furthermore, there are mutations that are particularly "troublesome" or completely lethal without affecting splicing! For instance, some muscular distrophies are caused by a single amino-acid replacement in a protein.

Interesting, But Not The First

[Anonymous Coward]

This is an interesting program, but it's hardly the first (coming from a grad student in Bioinformatics studying variation in the cancer genome). There are a large number of programs attempting to determine the impact of variation in the genome. We use these in our work with the exome (enrichment and sequencing of just the known coding elements in the human genome). A much larger challenge is finding true positive base and indel variation from next-generation sequencing, where alignment and sequencing artifacts lead to many false positive signals.

Here are two of the most popular:
SiFT: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC168916/
PolyPhen: http://www.ncbi.nlm.nih.gov/pubmed/12202775?dopt=Abstract

-Lee

okay...

[CosaNostra Pizza Inc]

So is this another tool the insurance companies could potentially use for genetic profiling...and at no additional cost to them? :/

Re:Superpowers

[Samantha Wright]

It's not really a simulation; it just tells you if a given sequence appears to resemble a splice site, and the probability that the single-nucleotide mutation specified by the user will alter that site's tendency to splice. Not a new accomplishment in biology by any means; the headline is sadly and totally overblown.

Re:Nothing new here.

[Samantha Wright]

Of course it only uses RefSeq! Don't you remember what your third-year professors taught you about the uncurated GenBank pool? Wikipedia users and terrorists post in there!

...actually, I'm pretty sure the only usage of RefSeq data is to establish the correct profile of what a splice site should look like. You can still feed it any FASTA sequence.

Re:It's begun

[Samantha Wright]

Nope. Not this story. Come back in five years, maybe. This is just an incrementally better method of doing a test we've had computers doing since the early nineties... and that we've been doing by hand for (close to) twice that period of time.