protein consensus sequence recognition site prediction software

[Hiro Protagonist]

Dear DIY-Biologists

I currently work on a project which aims to re-engineer TOR in S.cerevisiae BY4742 in order to lose its specifity for certain substrates.
As TOR1 [http://www.yeastgenome.org/cgi-bin/FUNGI/getSeq.pl?seq=YJR066W_BY4742] is a kinase which phosphorylates one of my proteins of interest - ATG13 [http://www.yeastgenome.org/cgi-bin/FUNGI/getSeq.pl?seq=YPR185W_BY4742],
I tried to figure out which sites in ATG13 are subject to TOR dependent phosphorylation and important for inhibitition of autophagy. This lead me to the ATG13 8SA mutant[http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2815578/] wherein eight important sites were identfied.

This again leads me to my current main problem, which is to identify where TOR is able to bind and/or identify (to) the related motif in ATG13 [S-X-S*-P].

Are there any programs available to predict these recognition sites [preferable with fasta support (not only pdb)] ?


Thanks in advance

Bruno Lederer

[Bryan Jones]

I'm not sure if this could help you, but I just released a program that might be of use. It's primarily designed to identify places where your protein differs from the consensus, so you can mutate it to increase the stability. But it also gives you outputs of an overall consensus sequence, and a spreadsheet of amino acid frequencies across the whole length of the protein. You can at least use that to tell you what parts of your protein are highly conserved, and see if any of those regions match the binding motif.
You can try it out at http://kazlab.umn.edu/

--
-- You received this message because you are subscribed to the Google Groups DIYbio group. To post to this group, send email to diy...@googlegroups.com. To unsubscribe from this group, send email to diybio+un...@googlegroups.com. For more options, visit this group at https://groups.google.com/d/forum/diybio?hl=en
Learn more at www.diybio.org
---
You received this message because you are subscribed to the Google Groups "DIYbio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to diybio+un...@googlegroups.com.
To post to this group, send email to diy...@googlegroups.com.
Visit this group at https://groups.google.com/group/diybio.
To view this discussion on the web visit https://groups.google.com/d/msgid/diybio/fa0ae295-eb5f-475f-ae37-e6866eee0eca%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Jeroen Delcour]

Uniprot is a great resource for these kind of things. Here's the page with all info about ATG13 in yeast: http://www.uniprot.org/uniprot/Q06628
Scroll down to "PTM / processing" and the "Amino acid modifications" and you'll find a list of known binding sites.

Hm, it seems all of them refer to the paper you've already found. But there's a lot of them it seems. Why do you want to find more? In my experience in-silico predictions are very unreliable, you'd still need to mutate and verify each possible site it finds anyway, which is a lot of work.

Cheers,
Jeroen

[Hiro Protagonist]

I think that I might have expressed what I am searching for in the wrong way.  I'm not searching for the sites in ATG13 but the parts in TOR which are able to bind to the sites. I found a program called COACH,  however this only shows drugs binding to it not proteins.

--

-- You received this message because you are subscribed to the Google Groups DIYbio group. To post to this group, send email to diy...@googlegroups.com. To unsubscribe from this group, send email to diybio+un...@googlegroups.com. For more options, visit this group at https://groups.google.com/d/forum/diybio?hl=en
Learn more at www.diybio.org
---

You received this message because you are subscribed to a topic in the Google Groups "DIYbio" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/diybio/aki0w3sGi5c/unsubscribe.
To unsubscribe from this group and all its topics, send an email to diybio+un...@googlegroups.com.

To post to this group, send email to diy...@googlegroups.com.
Visit this group at https://groups.google.com/group/diybio.

To view this discussion on the web visit https://groups.google.com/d/msgid/diybio/7bd6546f-f6d6-4e10-9192-94adf4f04dfa%40googlegroups.com.

[Xabier Vázquez Campos]

You have a list of resources in this paper:
http://www.sciencedirect.com/science/article/pii/S0959440X13001267

Good luck! That interaction is not detailed in any interaction database, all there is in BioGRID is quite useless other than saying that TOR1 phosphorylates ATG13
http://thebiogrid.org/98022/publication/tor-directly-controls-the-atg1-kinase-complex-to-regulate-autophagy.html

[Hiro Protagonist]

Thanks all

as i now have access to some predicted pdb files it would be interesting to know wether anyone knows a program where one is able to submit both involved proteins as input since predicted drugs binding to TOR1 are not as relevant as ATG13 itself as I already know that there is an interaction between those.

[Scott]

Hiro,

It helps to better understand the domain structure of TOR1 to focus on regions of interest. You can use Hidden Markov Models to see which parts of TOR1's amino acids map to domains found in other molecules. It is a statistical approach and sounds quite complex but practically it is quite simple to do, much like submitting an amino acid sequence for BLAST analysis. To do this you can search your FASTA TOR1 file against the PFAM database. I love this database since it shows the awesomeness of evolution - one molecule at a time! With all the genome data being pumped into GenBank the PFAM database automatically troll new protein sequences and incorporates those results in their dataset.

YJR066W_BY4742 827 986 827 986 PF11865.5 DUF3385 Family 1 161 161 209.1 3.0e-62 1 CL0020

YJR066W_BY4742 1461 1845 1461 1845 PF02259.20 FAT Family 1 345 392 432.9 8.3e-130 1 CL0020

YJR066W_BY4742 1952 2049 1952 2050 PF08771.8 Rapamycin_bind Domain 1 98 98 137.2 1.8e-40 1 No_clan

YJR066W_BY4742 2119 2366 2118 2368 PF00454.24 PI3_PI4_kinase Family 2 248 249 244.5 1.3e-72 1 CL0016

YJR066W_BY4742 2440 2470 2439 2470 PF02260.17 FATC Family 2 32 31 59.8 1.4e-16 1 No_clan

[Scott]

Dang, hit post prematurely!

I posted the PFAM search result of TOR1 in my previous post. PFAM correctly identifies the kinase domain and reveals a number of other domains. You can take the PFAM accession number (e.g. PF11865) and look that up to see how TOR1 compares to molecules with related domains. Domains that are marked DUF (e.g. DUF3385 ) are domains of unknown functions.

What is cool is that you can often clone these domains out and study them independently. This can help you to divide and conquer your molecule. You can, for example, use a yeast two hybrid system with the different domains to see what binds to wildtype ATG13 vs the mutant forms.

Hope this helps,

Scott

http://www.opensciencenet.org

[Scott]

And to address your last question have a look at VMD.

http://www.ks.uiuc.edu/Research/vmd/

I was trying to get proteins structures exported as STL files from VMD to display in Microsoft's HoloLens. Couldn't get the files uploaded to the HoloLens through OneDrive so I ended up blasting scorpions  and Robo- Fighters as they crawled out of the wall at my friend's place. A different type of fun, eh?

Cheers,

Scott

http://www.opensciencenet.org

[CodeWarrior]

I'm aware of http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html but I think you need a pdb in the database or to provide your own. If you don't have an experimentally proved protien folding have you looked at in silico fold prediction?

[Hiro Protagonist]

Thanks for the answers,

I already used PFAM earlier on to identify the catabolic domain, which actually was pretty helpful.
I used SPARKSX FOLD Recognition (http://sparks-lab.org/yueyang/server/SPARKS-X/) to gain Pdb models, however I am not quite aware of how trustworthy these predictions are.
I'm going to check out PISA and VMD later on, thanks again for the input ^^

I'm aware of http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html but I think you need a pdb in the database or to provide your own. If you don't have an experimentally proved protien folding have you looked at in silico fold prediction?

--
-- You received this message because you are subscribed to the Google Groups DIYbio group. To post to this group, send email to diy...@googlegroups.com. To unsubscribe from this group, send email to diybio+un...@googlegroups.com. For more options, visit this group at https://groups.google.com/d/forum/diybio?hl=en
Learn more at www.diybio.org
---
You received this message because you are subscribed to a topic in the Google Groups "DIYbio" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/diybio/aki0w3sGi5c/unsubscribe.
To unsubscribe from this group and all its topics, send an email to diybio+un...@googlegroups.com.
To post to this group, send email to diy...@googlegroups.com.
Visit this group at https://groups.google.com/group/diybio.

To view this discussion on the web visit https://groups.google.com/d/msgid/diybio/9b913177-5f6e-401b-a00c-014231e0f54b%40googlegroups.com.

[Josiah Zayner]

Hey Bruno,

     Unless there is an already determined interaction motif it is not very feasible to predict binding sites from amino acid sequences alone. Even with the protein structures the problem is extremely complicated because it is a three dimensional search. Most programs out there even if they do suggest they might be able to help have usually only been verified on extremely small and particular training sets.

     If you don't have a structure you are in bad starting spot. There are things like covariation analysis(looking for regions of the protein that vary or don't vary together) and trying to use that to predict spots that could be important in binding but still not great. If you haven't already you should create a homology model of the protein using something like SWISS Model(http://swissmodel.expasy.org/). Once you have that you can look for possible binding regions through analysis. Proteins tend to interact using loops or helices through salt bridges or hydrophobic surfaces. Use some or all of these servers to narrow things some maybe(http://rosettadesigngroup.com/blog/58/10-protein-protein-interface-prediction-servers/). After that the only real way to test is trial and error using site directed mutagenesis and mutating the protein and measuring binding between the protein and its ligand(binding partner).

     Molecular dynamics won't really help in this situation because the search space is so massive. You can't just throw two proteins(or even a protein and small molecule) into an MD simulation and hope they bind, they won't. Most MD simulations are for understanding thermodynamics and not kinetics(binding is a kinetic event though it involves thermodynamics) because without doing fancy techniques and extrapolating alot the timescale of MD simulation is in nanoseconds to low microsecond(even with the best super computers).

     If the binding effects an actual pathway that creates a phenotypic change you are doing great. Then you can just knock out the endogenous gene in yeast and and put it in a plasmid then using that system screen a bunch of mutants. If there is no great phenotypic change things become much more complicated. You would need to do binding assays, which most likely involves pull downs unless you have access to purified protein.

Hope this helps,
     Josiah Zayner
     http://the-odin.com

[Hiro Protagonist]

Thanks Josiah

you're right, it was probably kind of silly of me to think it would be possible to get that far without doing practical experiments.

In the beginning I already thought about performing a saturation mutagenesis, however I didnt have that much money and I thought it would be best to get as far as possible without using ressources. As I am backed by an Institute right now however I might actually be able to do such experiments.

The yeast is doing fine btw. ^^

To view this discussion on the web visit https://groups.google.com/d/msgid/diybio/5175bbe1-f865-40b9-982a-03110280ec98%40googlegroups.com.