RNA Sequence Assembly WEI Xueliang Overview • • • • • Sequence Assembly Current Method My Method RNA Assembly To Do Sequence Assembly • Goal : get the DNA/RNA sequence. • Machine cannot read whole genomes in one go, but rather small pieces between 20 and 1000 bases. • Define: Read = Tag = Fragment De novo sequence assembly Overview • • • • • Sequence Assembly Current Method My Method RNA Assembly To Do De novo sequence assembly Calculating the overlap need huge amount of time. DE BRUIJN GRAPH K-Mer : Length k substring of the Tag. Each nodes only have 4 out degrees at most. Hashing the node. “CTG”=>(132)4=(30)10 “CTG”=>”TGG” (132=)4 shift left. (1320)4 module (1000)4 (320)4 + (3)4 ‘G’ (323)4 DE BRUIJN GRAPH (CONT’) If there are repeats, like ”GACT” 3-Mer De Bruijn can not know which way is the correct way. 6-Mer can get the correct sequence. Larger K, better result. De novo sequence assembly Suppose use K = Length of Tag. (20-Mer) TGACGTAGCTATGTATTTTG GACGTAGCTATGTATTTTGT (no 20-Mer) Coverage is not enough to support large K. Overview • • • • • Sequence Assembly Current Method My Method RNA Assembly To Do MY METHOD. Tag length=6, K=3 When we have AAGACT? Try all the way: AAGACTC AAGACTT AAGACTG Check Tag : AGACTC The correct way should be AAGACTC Overview • • • • • Sequence Assembly Current Method My Method RNA Assembly To Do RNA ASSEMBLY ALTERNATIVE SPLICING The graph All cDNA sequences. RNA ASSEMBLY’S PROBLEM Merge? Index the sequence. RNA ASSEMBLY’S PROBLEM(CONT’) Solution? RNA ASSEMBLY’S PROBLEM(CONT’) Index Tags RNA ASSEMBLY’S PROBLEM(CONT’) Solution? Speed? SINGLE TAG’S LIMITATION |Yellow Sequence| >= Length of Tag Length of Tag 25-100bp. Single Tag is not enough! DATASET - PAIRED END TAGS Fragment length usually > 1k Some RNA sequence is shorter than 1k. TO DO Handle large data-sets. (10G) Improve accuracy. Using PETs data. Thanks!!