• Fragments
  • Fragment size
• Libraries
• Adaptors, vectors
• Reads
  • Read length
  • Single end
  • Paired end
  • Mate pairs

• Base calling
• Error rate
• Phred Quality
• Trimming by quality
• Clipping adaptors/vectors/tags
• Run

• Sanger
• Illumina
• 454
• SOLID
• PacBio
• nanoball
• Ion Torrent

• Run size
• Run price
• Read length
• Error rate

• de novo v/s reference based
• Overlay-layout-consensus
• Overlap
• Overlap Threshold
• Contigs
• Scaffolds
• Repeats

• Genome Length \(G\)
• Read Length \(L\)
• Number of reads \(N\)
• Overlap Threshold \(T\)
• Expected number of contigs

\[N\exp\left(-\frac{(L-T)N}{G}\right)\]

If \(T\) is too small, we risk confusing two parts of the genome

Before any biological considerations, the expected number of times we see a repeat of size \(T\) just by chance, is \(G/4^k\)

Thus, the probability that two reads match wrongly is \(4^{-T}\). And there are almost \(N^2\) pairs of reads

Expected number of chimeras: \(N^2 4^{-T}\)

• Depth
• Breadth of Coverage
• N50

• FASTA
• FASTQ
• SCF
• SAM
• BAM

• Cost of search without index
• Cost of search using an index
• Cost of making an index
• Space required to make an index