![Page 1: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/1.jpg)
Predicting Temperature-Sensitive Mutations
Chris PoultneyBonneau Lab
NYU
![Page 2: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/2.jpg)
2
● Find temperature-sensitive mutations● Method to generate “top 5” list● Use Rosetta to model mutations● Machine learning for prediction
Motivation
nativeranked predictions
0.91 C22T
0.88 S49P
0.87 I215N
![Page 3: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/3.jpg)
3
Conditional and “ts” Mutations
● Conditional Mutation– Wild-type (wt) phenotype under
permissive conditions– Mutant phenotype under
restrictive conditions● Temperature-Sensitive (ts)
– Restrictive condition is different temperature
● Context– Knock-out libraries (YKO)– Embryonic lethal phenotypeWikipedia: Genetics
Lyons LA, Imes DL, Rah HC, Grahn RA (2005) Tyrosinase mutations associated with Siamese and Burmese patterns in the domestic cat (Felis catus). Anim Genet 36: 119-126
![Page 4: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/4.jpg)
4
Causes of ts Behavior
● What changes at restrictive temperature?– Drop in level or activity of gene product
● Implications for Rosetta-based method– Likely to detect
● Decrease in stability● Failure to fold
– Unlikely to detect● Reduced function (e.g., catalysis, DNA binding)● Aggregation
Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan S, Varadarajan R (2004) Design of temperature-sensitive mutants solely from amino acid sequence. Proc Natl Acad Sci U S A 101: 7925-7930
Sandberg WS, Schlunk PM, Zabin HB, Terwilliger TC (1995) Relationship between in vivo activity and in vitro measures of function and stability of a protein. Biochemistry 34: 11970-11978
![Page 5: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/5.jpg)
5
Generating ts Mutations
Category Method(s) Screening Probability
Whole genome mutation EMS, X-ray, ... ~100,000 ~0.00001
Single gene mutation PCR mutagenesis ~10,000 ~0.0001
Single gene mutation Diploid shuffle ~10,000 ~0.0001*
Prediction from sequence Burial from sequence ~65 ~0.1
Prediction from structure Top5 ~5 ~0.4
Suzuki DT, Grigliatti T, Williamson R (1971) Temperature-sensitive mutations in Drosophila melanogaster. VII. A mutation (para-ts) causing reversible adult paralysis. Proc Natl Acad Sci U S A 68: 890-893Dohmen RJ, Wu P, Varshavsky A (1994) Heat-inducible degron: a method for constructing temperature-sensitive mutants. Science 263: 1273-1276Zeidler MP, Tan C, Bellaiche Y, Cherry S, Hader S, Gayko U, Perrimon N (2004) Temperature-sensitive control of protein activity by conditionally splicing inteins. Nat Biotechnol 22: 871-876Ben-Aroya S, Coombes C, Kwok T, O'Donnell KA, Boeke JD, Hieter P (2008) Toward a comprehensive temperature-sensitive mutant repository of the essential genes of Saccharomyces cerevisiae. Mol Cell 30: 248-258Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan S, Varadarajan R (2004) Design of temperature-sensitive mutants solely from amino acid sequence. Proc Natl Acad Sci U S A 101: 7925-7930
![Page 6: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/6.jpg)
6
Method Overview
native mutations ensembles
0,16.94,16.81,11.67,7.2,16.94,16.81,11.67,7.2,16.94,16.81,11.67,7.2,16.94,16.81,11.67,7.2,16.94,16.81,11.67,7.2,16.94,16.81,11.67,7.2,16.94,16.81,11.67,7.2,16.94,16.81,11.67,7.2,16.94,16.81,11.67,7.2,16.94,16.81,11.67,7.
features
prediction algorithm(classifier)
ranked predictions
0.91 C22T
0.88 S49P
0.87 I215N
?ts
non-ts
![Page 7: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/7.jpg)
7
Rosetta & Finding the “Sweet Spot”
● “Sweet Spot”: intermediate degree of destabilization– Moderate increase in energy (e.g.
fa_rep)● Proteins vary in starting energy
and properties (e.g. stability)● Rosetta score function● Allow structure to adjust to
mutation
green: ts blue: non-ts
![Page 8: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/8.jpg)
8
● Start with native structure● Model mutations at buried sites (<10% accessible)● Perform 50 relax runs
– Generates model ensembles and score files
Rosetta Protocol
![Page 9: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/9.jpg)
9
Relax Ensemble
relax.linuxgccrelease -database $MINI_DB -s YPL228W-W251A.pdb -native YPL228W.pdb -nstruct 50 -relax:fast -out:file:scorefile YPL228W-W251A.sc -out:pdb_gz
version: 3.0 release
![Page 10: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/10.jpg)
10
Scores to Features
● How to quantify effect of mutation?– Different starting energy– Different native qualities
● Requirements– Compare score term
distributions– Normalize across
proteins● Compare quartiles of
mutant and native ensembles
![Page 11: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/11.jpg)
11
Cele Scer Dmel0
10
20
30
40
50
60
70
80
90
100
Training Set Statistics
non-tsts
species
# sa
mpl
es
Machine Learning
● Input: 81 features– 75 from from Rosetta score
terms– 6 other (accessibility, etc.)
● Training set: yeast, worm, and fly from literature– 382 samples: 75 ts, 207 non-ts
● Algorithms– SVM-L: SVM with linear kernel– SVM-G: SVM with gaussian
kernel (RBF)?ts
non-ts
![Page 12: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/12.jpg)
12
Cross-Validation Results
SVM-L: 0.692±.074SVM-G: 0.808±.082rnd: 0.366
SVM-L: 0.359SVM-G: 0.113
![Page 13: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/13.jpg)
13
Top Features
● Most important features– 'Qn' suffix score-term-to-feature conversion– Residue change
● aminochange, p_aa_ppQ3, ramaQ2– Local structure change
● fa_repQ2, hbond_sr_bbQ2, hbond_bb_scQ1– Global structure change
● gdtmm2_2Q3, gdtmm1_1Q3– Changes within relax run
● Repack_stdev_scoreQ2, Repack_average_scoreQ2
![Page 14: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/14.jpg)
14
Experimental Validation
● Initial validation on three species– Worm: Kris Gunsalus & Fabio Piano– Fly: Claude Desplan– Yeast: David Gresham
● Current yeast validation– Made predictions on yeast actin
● 375 residues, well-characterized● Difficult to find ts mutations at random
– Chose 7 candidates from SMO-L, SVM-G top 5– Literature search: all mutations uncharacterized– [ insert cool results here ]
![Page 15: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/15.jpg)
15
Future Work
● Improving speed– Current algorithm: global (runs
over every residue)– In development: local (runs only
on residues near mutation)– Estimated speedup ~10-fold
● ts prediction for the masses– Public web server– Submit structure of interest– Receive ranked list of candidate
ts mutations
![Page 16: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/16.jpg)
16
Thanks
Dennis ShashaRich Bonneau
Glenn ButterfossKevin Drew
Kris GunsalusMichelle Gutwein
David JukamDavid Gresham
![Page 17: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/17.jpg)
17
Processing Scores:Quartile Method
● Calculate percentile of mutation ensemble quartiles Q1, Q2, Q3 w.r.t. native ensemble
Output :omegaQ1,omegaQ2 ,omegaQ3 =0.150 ,0.499, 0.672
![Page 18: Predicting Temperature-Sensitive Mutationsrosettadesigngroup.org/workshops/RCW2010/Presentations/poultney.pdf · Chakshusmathi G, Mondal K, Lakshmi GS, Singh G, Roy A, Ch RB, Madhusudhanan](https://reader030.vdocument.in/reader030/viewer/2022020416/5c783ea509d3f2cb498b9b46/html5/thumbnails/18.jpg)
18
Training Method
● Training Set– Split into 80% / 20%– Parameter selection: 10-fold CV on 80%– Testing: train on 80%, test on 20%– Repeat 5x
reduced training set (10x CV) test set
full training set
split 1
split 2
split 3
split 4
split 5