Mikhail Dvorkin Vladislav Isenbaev Eugene Kapun Genomics Quick

Mikhail Dvorkin Vladislav Isenbaev Eugene Kapun Genomics Quick Start Scientific advisors Acad. Konstantin Skryabin, Bioengineering RAS Prof. Anatoly Shalyto, SPb. SU ITMO

Collaboration with Bioengineering RAS • Bioengineering RAS – Conducts biological experiments – Sets problems – Provides biological data • SPb. SU ITMO – Develops algorithms and programs • Started in the end of 2009 • Why us? SPb. SU ITMO: Genomics Quick Start 2

SPb. SU ITMO at ACM ICPC Eugene Vladislav Kapun Isenbaev Georgiy Korneev Mikhail Dvorkin We train Zürich ETH May be, MIT? : -) SPb. SU ITMO: Genomics Quick Start 3

Mikhail Dvorkin SPb. SU ITMO: Genomics Quick Start 4

Eugene Vladislav Kapun Isenbaev SPb. SU ITMO: Genomics Quick Start 5

SPb. SU ITMO: Genomics Quick Start 6

Genome Team Coach • Georgiy Korneev Members • Mikhail Dvorkin • Vladislav Isenbaev • Eugene Kapun SPb. SU ITMO: Genomics Quick Start 7

Problems Being Solved • DNA assembly de novo based on pair reads – Generalized suffix tree traversal – Reduction to single reads • DNA alignment with transfers SPb. SU ITMO: Genomics Quick Start 8

DNA Assembly 1 Generalized suffix tree traversal

Suffix Tree • • Built upon reads Arc weight: number and quality of reads Possible extensions Erroneous nucleotides detection SPb. SU ITMO: Genomics Quick Start 10

Building up a Contig • Start with high-quality read • Use pair reads to select a nucleotide – “Backward” – match the past – “Forward” – match the future • Build up to a branch SPb. SU ITMO: Genomics Quick Start 11

Results • Caenorhabditis elegans • Escherichia coli K-12 Mean coverage 156 125 100 Contig length 100% 76% SPb. SU ITMO: Genomics Quick Start 12

DNA Assembly 2 Reduction to single reads

Concept • De Bruijn graph with all reads • Pair reads – Path in the graph – Low density – backtracking – Slow – Meet-in-the-middle SPb. SU ITMO: Genomics Quick Start 14

Error detection • Poorly covered vertices – Erroneous – Delete them – Repeat • Paths – Single reads – Use another tool SPb. SU ITMO: Genomics Quick Start 15

Results • 60% erroneous reads detected • < 0. 1% errors left after one iteration • 99. 5% DNA coverage SPb. SU ITMO: Genomics Quick Start 16

DNA Alignment with transfers

Concept • Parts • Matched (small edit distance) • Unmatched • Swapping allowed • Penalties • Number of parts • Edit distance in matched parts • Length of unmatched parts SPb. SU ITMO: Genomics Quick Start 18

Implementation First DNA • Tear into small pieces • Hash ‘em and store ‘em Second DNA • Tear into small pieces • Look them up • Build them up SPb. SU ITMO: Genomics Quick Start 19

Results Bacteria 1. NC_012759 2. NC_012947 3. NC_012967 4. NC_010067 5. NC_002952 1 2 3 4 5 100% 90% 88% 11% 0. 1% 90% 100% 96% 11% 0. 1% 89% 97% 100% 11% 0. 1% 11% 100% 0. 1% 100% SPb. SU ITMO: Genomics Quick Start 20