Bioinformatics

Henry WH Hack v2.0

6:56 AM

The least I can do to promote this site:

UBUNTU!

Yeah people promote Area51.SE proposals on there SE account it's normal.

Even this guy does it:

Mithrandir ♦, Tauremornalómë

5.5k 1 24 65

terdon

3:28 PM

@KamilSJaron thanks for the answer. I am afraid my question was not very clear. I am referring to the bwa-backtrack, I just hadn't realized that the aln+samse+sampe was called backtrack. Stupid, I know.

Kamil S Jaron

3:50 PM

@terdon No worries. I find stupider that I did not know that samse+sampe is backtrack. What is read length? Is it <100 ?

terdon

@KamilSJaron I don't see why, I'm the guy asking the question :)

And no, these are regular Illumina reads, the two samples I mentioned are very very well known standards and used to validate pipelines all over the place.

Gimme a sec and I'll check the read length more specifically.

Kamil S Jaron

@terdon for some reason I believe that the one who is trying to answer should know bit better..

@terdon We have also uploaded fastq and bam files from ~300x total coverage of 150x150bp HiSeq2500 sequencing

found it

terdon

@KamilSJaron Only if the question is clear enough. The onus is on the asker to make what they are asking about clear.

Kamil S Jaron

Than it is even stranger..

backtrack is designed for older generation of high throughput Illumina sequencers (for reads < 70 I think...)

"The first algorithm is designed for Illumina sequence reads up to 100bp"

How comes that it performes better on 150bp reads... That sounds creepy.

We need Heng Li to join our community!

terdon

Heh, yes.

Kamil S Jaron

3:56 PM

@terdon I guess if you would make a blog post and shared it on twitter, that he would reply.

terdon

I was also surprised. I ran the benchmarks back when I fisrt started working here in december 2015 and I don't have the data anymore. But yes, using bwa-mem gave worse results on these (and other) samples than backtrack.

@KamilSJaron I could send him an email too but I don't see why he's the person to ask. I mean, he's already claimed it's better, he doesn't have anything else to add.

Kamil S Jaron

@terdon He is person who wrote both those algorithms, if anybody could figured out why in this case the results were like this, it will be him.

terdon

Yeah, the vast majority are at ~150nt length.

@KamilSJaron Maybe. But anyone who's worked with them and run benchmarks could also have useful input.

I really should run some more benchmarks and collect the data, I guess.

Kamil S Jaron

@terdon It would be interesting to try on different datasets as well just to see if it is a special case or general property.

@terdon I can try to run it on my stick insects...

terdon

Ooh, you're working on stick insects?

Cool! Those things are funky :)

But why? it's a strange model organism.

Kamil S Jaron

4:04 PM

I study sexual reproduction modes and stick insects have multiple transitions from sexual to asexual reproduction. So we sequence them all.

terdon

Are they related? The various types and species of stick insects, I mean. Are they close?

Kamil S Jaron

I work only with genus Timema which has a common ancestor ~30 mya.

The asexual lineages always diverged from the sexual ancestors from 0.5 - 2 mya.

There are at least 7 independent transitions in the genus (and we sequence 5 sister species)

terdon

And even within the same genus, I imagine the phenotypic variation to be enormous, right? Depending on habitat, presumably.

Kamil S Jaron

To get an idea you can take a look at https://en.wikipedia.org/wiki/Timema#phylogeny
but this is not the latest phylogeny (as you can imagine)

No, they have almost exactly same shape, just different colours.

terdon

Ah, OK, I had sort of thought that the phenotypic changes would be more superficial.

Meaning that the underlying genome wouldn't be all that different. Not really sure why I'd assume so, come to think of it.

Oh hey, is their genome published?

Do any of them have selenoproteins?

Selenoprotein

In molecular biology a selenoprotein is any protein that includes a selenocysteine (Sec, U, Se-Cys) amino acid residue. Among functionally characterized selenoproteins are five glutathione peroxidases (GPX) and three thioredoxin reductases, (TrxR/TXNRD) which both contain only one Sec. Selenoprotein P is the most common selenoprotein found in the plasma. It is unusual because in humans it contains 10 Sec residues, which are split into two domains, a longer N-terminal domain that contains 1 Sec, and a shorter C-terminal domain that contains 9 Sec. The longer N-terminal domain is likely an enzymatic...

Kamil S Jaron

4:18 PM

@terdon only one of them is published. I have assembled the rest couple of weeks ago..

@terdon No idea, why it is interesting to look at it?

@terdon The age of the genus is bit more than for instance primates (to give a scale)

terdon

@KamilSJaron Well, because selenoproteins were thought to be essential for animals, but then we found that insects had lost them. Or, rather, that many insects had lost them. Some drosophilas have them, others do not. Most dipterans seem to have them, some but most other insect lineages do not. And the phasmids have never been checked.

I did my PhD on these things (I found this loss in insects) and so I was curious. If it's been published, I'll ask a friend who's still working on them. Just wondering what the situation would be in this genus.

Kamil S Jaron

@terdon That's cool! Stick insects are basal insect clade..

terdon

@KamilSJaron It's been a while, I don't remember the details. Where would they fit in the image above? Are they in one of the endopterygota, paraneoptera or orthopteroida or would they be in a separate branch?

Kamil S Jaron

@terdon I think I can provide early access we willbe very explicit about what is allowed and what it is not...

terdon

Nah, it's OK, I don't have time to look into this and I haven't touched anything of the sort in the last 10 years. This is just idle curiosity. If you know off the top of your head, then great. If not, no worries. Thanks though!

Maybe I'll email you next time I'm in Lausanne and I can come have a look :P

(the startup I work for is based in the EPFL)

Kamil S Jaron

4:28 PM

@terdon Orthopteroidea

@terdon hehe, you have two years to do so...

terdon

Ah! So I would expect them to have it. Cool. Which one is the publicly available genome?

@KamilSJaron :P

Kamil S Jaron

5:09 PM

@terdon Timema californicum (Stick Insect Genomes Reveal Natural Selection’s Role in Parallel Speciation)

terdon

Cool, thanks

terdon

5:28 PM

Heh, cool. Looks like it doesn't have them either. Well, that's what a couple of blast searches suggest anyway.

Kamil S Jaron

@terdon u, that sounds interesting...

terdon

Yeah, selenoproteins are really interesting. I like all exceptions and this one is a good one. For example, c. elegans has kept the whole system (~5 separate factors) only to encode a single selenocysteine residue.

So that's one single UGA codon recoded in the entire genome and yet the whole complex machinery is kept around for it.

Kamil S Jaron

@terdon I will get back to you once I will read your paper Charles...

terdon

Ah, you found me out eh? :P

Kamil S Jaron

@terdon was not that difficult..

terdon

5:37 PM

I'm sure it wasn't.

Not when talking about things like this.

Kamil S Jaron

@terdon Google scholar : selenoprotein arthoropoda -> the first hit

terdon

Yeah, it's a tiny field. I used to enjoy telling people I'm one of the top 5 experts in the world in my field. I could always conveniently leave out the fact that there are only 5 people in the world working on it so I'm one of the top 5 by definition :)

Kamil S Jaron

@terdon can not happen to me, too many people study sex..

terdon

True.

Transcript for