Information-Theoretic Analysis of Molecular (Co)Evolution Using - - PowerPoint PPT Presentation

Information-Theoretic Analysis

• f

Molecular (Co)Evolution Using Graphics Processing Units

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 1

. . . AEERYAEYKEAFTLFDSDGD. . . . . . TEEQGRQFRQM FEM

• FDKNGD. . .

. . . TDEQQRQYRQM

• FETFDKDGN. . .

. . . TKEQVEEFKQAFSM

• FDTDGD. . .

. . . SEEQVAEFKEAFDRFDKNKD. . . . . . SKEQVAKFKEAFDRI DKNKD. . . . . . SPEQVAEFKQAFSRFDKNGD. . . . . . SEEQVAKFKAAFSRFDTNGD. . . . . . PPEQVAKFKEVFSRFDKNGD. . .

Michael Waechter, Kathrin Jaeger, Stephanie Weissgraeber, Sven Widmer, Michael Goesele, and Kay Hamacher

. . . AEERYAEYKEAFTLFDSDGD. . . . . . TEEQGRQFRQM FEM

• FDKNGD. . .

. . . TDEQQRQYRQM

• FETFDKDGN. . .

. . . TKEQVEEFKQAFSM

• FDTDGD. . .

. . . SEEQVAEFKEAFDRFDKNKD. . . . . . SKEQVAKFKEAFDRI DKNKD. . . . . . SPEQVAEFKQAFSRFDKNGD. . . . . . SEEQVAKFKAAFSRFDTNGD. . . . . . PPEQVAKFKEVFSRFDKNGD. . .

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 2

• Huge amount of Multiple Sequence Alignments (MSAs)

available, some of them really large

• E.g., HIV protease [1]:

> 45,000 sequences of length > 1400

• Put them to use for coevolutionary and structural analysis
• But: Our computations take >25 days

[1] Pan et. al.:“The HIV positive selection mutation database”

Motivation

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 3

• In this talk we will show…
• MSA analysis using Mutual Information
• GPU parallelization & speed improvements
• 3-point Mutual Information
• an application to a well-known protein
• that the use of this is beneficial

Outline

contributions

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 4

• Given an MSA:
• Mutual Information between two columns

(correlation  coevolution):

• Iteration over all column pairs  MI matrix:

Introduction – Mutual Information

Sequence 1: AEERYAEYKEAFTLFDSDGD. . . Sequence 2: TEEQGRQFRQM FEM

• FDKNGD. . .

Sequence 3: TDEQQRQYRQM

• FETFDKDGN. . .

Sequence 4: TKEQVEEFKQAFSM

• FDTDGD. . .

Sequence 5: SEEQVAEFKEAFDRFDKNKD. . . Sequence 6: SKEQVAKFKEAFDRI DKNKD. . . Sequence 7: SPEQVAEFKQAFSRFDKNGD. . . Sequence 8: SEEQVAKFKAAFSRFDTNGD. . .

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber

• MI is sensitive to underlying amino acid distribution
• Computational Normalization: Shuffling Null-Model [2]
• Is MI distinguishable from “random evolution” MI?

[2] K. Hamacher: “Relating sequence evolution of HIV1-protease to its underlying molecular mechanics”

Introduction – Shuffling Null-Model

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 6

• Compute original MI
• Iterate 10,000 times:
• Shuffle each MSA column
• Compute rand. MI matrix
• Normalize original MI

using random MI:

Introduction – Shuffling Null-Model

6

• AEER. . .
• TEEQ. . .
• TDEQ. . .
• SEEQ. . .
• SKEQ. . .
• PPEQ. . .
• SEEQ. . .
• SEEQ. . .
• PEEQ. . .
• TPEQ. . .
• AKEQ. . .
• TDER. . .
• SEEQ. . .
• TDER. . .
• TKEQ. . .
• SEEQ. . .
• APEQ. . .
• PEEQ. . .

. . . . . .

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 7

• Highly compute intensive
• HIV-1 protease on single core:
• MI computation for all column pairs: ~3.5 min
• Repeat for 10,000 iterations: > 25 days
• But:
• Computation of each MI matrix entry independent of all others
• Shuffling of each MSA column independent of all others
• Parallelizable (to hundreds of thousands of threads)

Massive parallelism

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 8

• Iterate 10,000 times:
• Shuffling

– Map MSA columns to blocks of threads – Shuffle columns (GPU suited algorithm) – Synchronize

• MI computation

– Map MI matrix entries to blocks of threads (suitable for MSA access pattern) – Compute MI matrix entries – Synchronize

• Combine results & normalize orig. MI with randomized MI

GPU Implementation

. . . AEERYA. . . . . . TEEQGR. . . . . . TDEQQR. . . . . . TKEQVE. . . . . . SEEQVA. . . . . . SKEQVA. . . . . . SPEQVA. . .

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 9

• Problem size dependent

Speed Results

GeForce GTX 480 4 threads

• n Core i7‐960

Calmodulin 753 sequences

• f length 264

1.1 min 13.4 min ~ 12x speed‐up HIV‐1 protease > 45,000 seqs.

• f length > 1400

1.85 days 7.3 days ~ 4x speed‐up

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 10

• One order of magnitude speed-up
• Quickly redo previous steps (e.g., alignment) and recompute

MI

• New analysis tool feasible:

3-point MI: Coevolution of a ‘3-clique’ of MSA columns

• Can we deduce more information from 3-point MI than we

could from 2-point MI alone?

Implications

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 11

• 149 amino acids
• Ca2+ binding

 conformational change

• Regulates various signaling pathways

Calmodulin

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 12

• Finding coevolving pairs of

amino acids

• Structural or functional

connection

• Here: Coevolution within N-

and C-terminus

• Ca2+ binding
• Propagation of conformational

change

• Conserved inner helix
• No coevolution without variation

Coevolution in Calmodulin – 2-point MI

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 13

• ‘3-cliques’ of amino acids
• Higher order correlations
• Concerted motions
• Binding sites

Coevolution in Calmodulin – 3-point MI

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 14

• ‘3-cliques’ of amino acids
• Color indicates the frequency

with which an amino acid contributes to the ‘3-cliques’ set

• Key residues for important

functions

Coevolution in Calmodulin – 3-point MI

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 15

• MI for coevolutionary analysis
• GPU implementation ~10x faster on typical MSAs
• 3-point MI analysis possible in acceptable time
• 3-point MI does reveal new insights
• Next step could be k-point MI
• It may be possible to detect key residues in unknown proteins

Conclusions

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 16

• Multi-GPU parallelization:
• Distribute Shuffling Null-Model iterations among GPUs
• First tests: 32 GPUs  ~32x speed-up (on top of basic GPU speed-

up!)

What happened since?

June 18, 2012 | ECMLS 2012 | Michael Waechter & Stephanie Weissgraeber 17

Thank you.

Please visit tinyurl.com/tud‐comic for code & documentation

• r contact us.