1. Identification and evaluation of causative genetic variants corresponding to a certain phenotype Xidan Li

2. Outline SIT- identify and evaluate the causative genetic variants within a QTL/GWAS defined region. PASE - evaluate the effect of amino acid substitution to the hosting protein function DIPT - to identify causative genes underlying an expression phenotype Parallelizing computing

3. Genetic variances identification

4. Possible solutions?

6. Working process of SIT VCF file SNPs analysis in non-coding regions SNPs analysis in coding regions Splicing sites CpG island UTR region Non-synonymous SNPs PASE Candidate genes with candidate SNPs List of ranking Non- synonymous SNPs Ensembl

7. Sample results

8. Non-synonymous SNPs are ranked

9. The life is easy!

10. Amino acid substitutions effects prediction

11. Effect of amino acid substitutions

13. Selected seven physico-chemical properties of Amino acids Seven Physiochemical properties of Amino acid Transfer free energy from octanol to water Normalized van der Waals volume Isoelectric point Polarity Normalized frequency of alpha-helix Free energy of solution in water Normalized frequency of turn

14. Formula for conservation calculation 1-.95 N Probability of 20 different AAs in a position for N random equal frequent sequences. n observed /N total (1-.95 N )*(n observed /N total ) Blast search clustalw

15. Protein kinase AMP-activated gamma 3 (PRKAG3) gene (R200Q) in AMPK  3 in purebred Hampshire pigs – RN (V199I) in AMPK  3 Co-participate in the effective process with R200Q RN that causes excess glycogen content in pig skeletal muscle Milan D, et. al. (2000). A mutation in PRKAG3 associated with excess glycogen content in pig skeletal muscle. Science 288 (5469): 1248–51. Ciobanu,D, et. al. (2001). Evidence for New Alleles in the Protein Kinase Adenosine Monophosphate-Activated 3-Subunit Gene Associated With Low Glycogen Content in Pig Skeletal Muscle and Improved Meat Quality. Genetics, 159, 1151-1162.

16. Genes IDCoordinateREFALTConservations score (MSAC) PASE score PASEC (combined) score PRKAG_3200RQ0.930.540.50 PRKAG_3199VI0.850.140.12 (R200Q) Cause major increase in the muscle glycogen content (V199I) Contribute with smaller effect Ciobanu,D, et. al. (2001). Evidence for New Alleles in the Protein Kinase Adenosine Monophosphate-Activated 3-Subunit Gene Associated With Low Glycogen Content in Pig Skeletal Muscle and Improved Meat Quality. Genetics, 159, 1151-1162.

17. Testing with SIFT and POLYPHEN Conservation scores (MSAC) PASE scores (Physico- chemical properties changings) PASEC score (combined) SIFT Tolerated (1987)0.470.390.18 Deleterious (1351)0.600.510.30 PolyPhen Benign (1637)0.440.370.16 Possibly damaging (539) 0.560.430.24 Probably damaging (1162) 0.630.530.33

18. Features Other tool SIFT, PolyPhen MAINLY rely on calculating sequence conservation scores (finding homologous sequences). PASE not only uses the physico-chemical property changing score, but also combine with sequence conservation score Potentially being able to analyze the evolutionary-distant protein sequence

20. From expression phenotype to association genotype

22. Sample result of DIPT

24. www.computationalgenetics.se/DIPT/

25. Parallelizing computing

26. Principle of parallelizing computing

27. Multiple threads – efficient work

28. Single thread - tough job!

29. Usually in the loop Data must be independent

30. GPU vs. CPU

32. Cuda Vs. C #include // Prototypes __global__ void helloWorld(char*); // Host function int main(int argc, char** argv) { int i; // desired output char str[] = "Hello World!"; // mangle contents of output ; the null character is left intact for simplicity for(i = 0; i < 12; i++) str[i] -= i; // allocate memory on the device char *d_str; size_t size = sizeof(str); cudaMalloc((void**)&d_str, size); // copy the string to the device cudaMemcpy(d_str, str, size, cudaMemcpyHostToDevice); // set the grid and block sizes dim3 dimGrid(2); // one block per word dim3 dimBlock(6); // one thread per character // invoke the kernel helloWorld >>(d_str); // retrieve the results from the device cudaMemcpy(str, d_str, size, cudaMemcpyDeviceToHost); // free up the allocated memory on the device cudaFree(d_str); // everyone's favorite part printf("%s\n", str); return 0; } // Device kernel __global__ void helloWorld(char* str) { // determine where in the thread grid we are int idx = blockIdx.x * blockDim.x + threadIdx.x; // unmangle output str[idx] += idx; } #include int main(void) { printf("Hello World\n"); return 0; }

33. Thank You!