ou supercomputer symposium – september 23, 2015 parallel programming in the classroom analysis of...

69
OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming Seminar CS4973 Spring 2015

Upload: sophie-griffith

Post on 18-Jan-2018

213 views

Category:

Documents


0 download

DESCRIPTION

OU Supercomputer Symposium – September 23, 2015 We Started Here... Early Henry!

TRANSCRIPT

Page 1: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Parallel Programmingin the Classroom

Analysis of Genome Data

Karl Frinkle - Mike MorrisParallel Programming Seminar CS4973

Spring 2015

Page 2: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

We Started Here . . .

Page 3: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

We Started Here . . .Early

Henry!

Page 4: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Our first cluster – made from junk – but it

worked!

We Progressed To . . .

Page 5: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Then Found LittleFe . . .

Page 6: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Next We Used Sooner . . .

Page 7: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Now We Have . . .

Page 8: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Analyzing Genome Data

Page 9: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

PHASE 1 – Write Code

Page 10: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

PHASE 1 – Write CodeDefinitions:

SNP: single-nucleotide polymorphismpronounced “snip” is a DNA sequence commonly varying within a population

Page 11: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

PHASE 1 – Write CodeDefinitions:

SNP: single-nucleotide polymorphism

rsid: Reference SNP cluster ID

pronounced “snip” is a DNA sequence commonly varying within a population

access number used to refer to specific SNPs

Page 12: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

• Harvard PGP* Database• 23andME

* PGP - Personal Genome Project

# rsid chromosome position genotype. . .rs12564807 1 734462 AArs3131972 1 752721 GGrs148828841 1 760998 CCrs12124819 1 776546 AA . . .

PHASE 1 – Write Code

Page 13: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Harvard PGP DatabasePer person, there were about 1,000,000 snips.

PHASE 1 – Write Code

# rsid chromosome position genotype. . .rs12564807 1 734462 AArs3131972 1 752721 GGrs148828841 1 760998 CCrs12124819 1 776546 AA . . .

Page 14: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

• We started with 200 profiles• Then gave ‘em names• That was about 5G of data

Per person, there were about 1,000,000 snips.

PHASE 1 – Write Code

# rsid chromosome position genotype. . .rs12564807 1 734462 AArs3131972 1 752721 GGrs148828841 1 760998 CCrs12124819 1 776546 AA . . .

Page 15: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

15

!! Important Info for HPC Instructors !!

• 5G of data isn’t necessarily “big”• Our clusters are significantly small

Page 16: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

16

!! Important Info for HPC Instructors !!

• 5G of data isn’t necessarily “big”• Our clusters are significantly small

• We’re teaching concept and techniques

• We can easily scale up to Boomer

Page 17: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

• Search for a particular rsid for a given person• Ditto for many persons• Both of the above for a collection of rsids• Compare 2 persons’ makeup• we used a sliding window algorithm

• Compare many persons’ makeup

PHASE 1 – Write CodeSeveral programs begged to be written, andall were great candidates for parallelization.

Page 18: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .PENELOPE KARDASHIANi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AGi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs2340592 1 910935 --rs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 CT

STONEY BURKEi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AAi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs2340592 1 910935 GGrs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 TT

Page 19: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .PENELOPE KARDASHIANi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AGi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs2340592 1 910935 --rs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 CT

STONEY BURKEi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AAi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs2340592 1 910935 GGrs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 TT

If one person’s rsid was unrecorded, both were

tossed out.

Page 20: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .PENELOPE KARDASHIANi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AGi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 CT

STONEY BURKEi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AAi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 TT

FIRSTVARIANCE

Page 21: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .PENELOPE KARDASHIANi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AGi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 CT

STONEY BURKEi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AAi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 TT

VARIANCES

Page 22: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .PENELOPE KARDASHIANi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AGi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 CT

STONEY BURKEi6019305 1 891343 GGrs13303106 1 891945 AGi6019306 1 894379 GGrs13303010 1 894573 AAi6019308 1 897792 CCi6019309 1 898082 AArs6696281 1 903104 CTi6019310 1 905681 CCi6019311 1 906114 CCi6019312 1 907666 AAi6060381 1 909238 CGrs13303118 1 918384 GTrs78164078 1 921071 GGrs6665000 1 924898 ACrs2341362 1 927309 CCrs9777703 1 928836 TT

VARIANCESA set number of variances

were allowed in a given “block” size w/o causing a

“difference”.

Page 23: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .------@@------------------@@--@@----@@--@@------------

Page 24: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .This represents rsid

locations. “@@” means difference.

“—” means agreement.

------@@------------------@@--@@----@@--@@------------

Page 25: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .This represents rsid

locations. “@@” means difference.

“—” means agreement.

For this example, if 2 or more do not agree in a group of 8, then record a difference.

Otherwise it is recorded as a match.

------@@------------------@@--@@----@@--@@------------

Page 26: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .This represents rsid

locations. “@@” means difference.

“—” means agreement.

For this example, if 2 or more do not agree in a group of 8, then record a difference.

Otherwise it is recorded as a match.

------@@------------------@@--@@----@@--@@------------

It’s a “best case” algorithm. If an rsid is ever in a match box of 8, then it is forever a match.

Page 27: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Page 28: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 29: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 30: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 31: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 32: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 33: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 34: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 35: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Green = MatchRed = Difference

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Page 36: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

It’s a “best case” algorithm. If an rsid is ever in a match box of 8, then it is forever a match.

Page 37: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

It’s a “best case” algorithm. If an rsid is ever in a match box of 8, then it is forever a match.

Page 38: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 39: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 40: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 41: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 42: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 43: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 44: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 45: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@------------

Green = MatchRed = Difference

Page 46: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@--------------

Green = MatchRed = Difference

Page 47: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@--------------

Green = MatchRed = Difference

Page 48: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@--------------

Green = MatchRed = Difference

Page 49: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Sliding Window Technique . . .For this example,

if 2 or more do not agree in a

group of 8, then record a

difference. Otherwise it is recorded as a

match.

------@@------------------@@--@@----@@--@@--------------

Green = MatchRed = Difference

Page 50: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

PHASE 2 – Web GUI

Page 51: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

PHASE 2 – Web GUI

Page 52: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

PHASE 2 – Web GUI

Currently we have 5 options.

Page 53: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Single Person

Page 54: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Single Person

Page 55: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Multiple Person

Page 56: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Multiple Person

Page 57: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Group Allele

Page 58: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Group Allele

Page 59: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Two Person Compare

Page 60: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Two Person Compare

Page 61: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Two Person Compare

Very different – no connection.

(Blue is no match.)

Page 62: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – Two Person Compare

One-allele search.Blue is no match,green is match.

Page 63: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – System Monitor

Page 64: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – System Monitor

Group allele – accesses all files on all nodes.

Seeking 2 alleles.

Page 65: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – System Monitor

Group allele – accesses all files on all nodes.

Seeking 2 alleles.

Head node doing traffic only – not relatively busy.

Page 66: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – System Monitor

Group allele – accesses all files on all nodes.

Seeking 6 alleles.

Head node doing traffic only – not relatively busy.

Page 67: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – System MonitorTwo-person compare,

files on same node.

Page 68: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Web GUI – System MonitorTwo-person compare,

files on different nodes.

Page 69: OU Supercomputer Symposium – September 23, 2015 Parallel Programming in the Classroom Analysis of Genome Data Karl Frinkle - Mike Morris Parallel Programming

OU Supercomputer Symposium – September 23, 2015

Thank You!

We especially thank Henry Neeman, Charlie Peck, Tom Murphy, and everyone else associated with OU IT, the

LittleFe project, the SOSU IT guys and all of our colleagues and friends in the educational community involved with

HPC, for all the help we have received.

Karl FrinkleMike Morris