1
2
3
4
5	"Clone the DNA" Clone the DNA. Generate a ladder of labeled (colored) molecules that are different by 1 nucleotide. Separate mixture on some matrix. Detect fluorochrome by laser. Interpret peaks as string of DNA. Strings are 500 to 1,000 letters long 1 machine generates 57,000 nucleotides/run Assemble all strings into a “whole”.
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41	AB SOLiD: Dibase Sequencing
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66	"Alignment to genome" Alignment to genome Computationally expensive It is never a good idea to simply align RNA-seq data to the genome Need a spliced aligner or a surrogate (such as including exon-exon junction sequences in ‘genome’) Alignment to transcriptome Reads deriving from non-genic structures may be ‘forcibly’ (and erroneously) aligned to genes Incorrect gene expression values False positive SNVs Many other potential problems Assembly Low expression = difficult/impossible to assemble Misassemblies/fragmented contigs due to repeats Requires vast amounts of memory
67
68
69
70	"Clean up" Clean up ↓ Clustering ↓ ncRNA matching (Rfam V10 ) ↓ Known miRNA Matching (miRBase V16) ↓ Comparative miRNAomics
71
72
73
74	The “old” biology
75	The “new” biology
76
77	"The speed of your CPU..." The speed of your CPU determines how quickly it can process instructions Many bioinformatics operations fall into the “embarrassingly parallel” category Getting a results faster is as simple as adding more CPUs => clusters Though clusters still dominate, increasing attention is being paid to servers with large number of cores
78	"Typical sizing is 2GB of..." Typical sizing is 2GB of RAM per core This works fine for most aligners Certain aligners require a minimum number of cores for optimal efficiency 16-48GB RAM per node Assemblers typically need much more RAM If you don’t have enough RAM, the CPU will need to make use of the disk storage – When a computer has run out of RAM it is said to be “swapping” Some aligners have a minimum memory limit for optimal performance Human de novo requires lots of RAM!
79	"Sequencers generate a lot of..." Sequencers generate a lot of data Including quality values and additional files (alignments etc) create multiplier Binary formats such as BAM are helping, but there are limited binary formats for basic data Life and Illumina are both moving to binary formats for basic data Debates on what needs to be stored RAID Scaling
80
81
82