Anchor length

In TopHat, the minimum number of bases that need to be on either side of a junction for it to be reported: 8.

The probability of an A is 0.4, T is 0.4, C is 0.1, and G is 0.1 We can score A or T as 0.4/25 = 1.6, and G or C as 0.1/0.25 = 0.4. Calculated individually.

P ("AGCTAT" | M) = (0.1)²(0.4)⁴ = 0.000256.

Aligner

Software which does the first step of RNA sequence analysis: it aligns reads. Takes unmapped reads and attempts to find a genomic position for each read. Some reads will align across introns, and some reads will be polymorphic to the genome, caused by indels and SNPs. For a DNA sequence 100 bp long, the probability that a random 4-nucleotide sequence will align is 6.22315 x 10^-61; a very small probability. One sequence may align to several sites due to gene duplications and translocations. A good alignment has very similar or identical sequences. If you are comparing an RNA sequence to the sequenced genome of the same organism, There should be a perfect alignment. If you are comparing to a reference genome, there will be around 1% of natural variation in humans. If there are too many SNPs and indels, the read won't align; this judgement is made by the software. Millions of sequenced DNA fragments are aligned, typically to a reference genome of the same species.

Derivatives of score ratios have been calculated for use in the context of sequence alignment of amino acids. The ratio can give the probability of two amino acid residues being aligned in two homologous sequences (p_H) divided by the probability of two amino acids being aligned in two unrelated sequencdes (q_R). Given a known set of homologous sequences, we can calculate p_H. Given the frequency of amino acids, we can calcualte q_R. We can score each amino acid.

S = log (p_H / q_R)

.gff file

Optional to use with alignment software. Tells you what the genes are.

The probability of an A is 0.1, T is 0.1, C is 0.4, and G is 0.4. Used for thermophilic bacteria, where CG is enriched.

P ("AGCTAT" | M) = (0.1)⁴(0.4)² = 0.000016

Chromatin ImmunoPrecipitation sequencing

A method of sequence analysis that looks for associations between DNA-binding proteins and the DNA sequences they bind to. This method is often used to analyze the effects of transcription factors on specific regions of DNA. This offers the exact location and sequence to which the regulatory factor binds, which can then be used for further analysis of these regulatory factors and their effects on gene expression. Combines chromatin immunoprecipitation with high-throughput DNA sequencing.

1. Cross-linking with formaldehyde

2. Shearing via cell disruption and sonication or enzymatic digestion

3. Immunoprecipitation and enrichement

4. Sequencing

Software which estimates gene expression and expression differences. Estimates gene and gene isoform expression, and does a pairwise comparison between samples to find significant changes in transcript abundance, splicing, and promoter use. Output files include FPKM tracking files, count tracking files, and read group tracking files. Differential splicing and promoter use is only calculated if the number of replicates is sufficient. Based on FPKM, compares the log-ratio of the gene's expression two conditions against 0.

Y = (FPKM_a / FPKM_b)

Abundance output

Reports abundance as FPKM. Normalizes for transcript length and library size.

In eukaryotes it is often methylated in a CG context, except around genes. Methylation occurs outside promoter regions is mutagenic, changing C into T. Using these regions, we can calculate transition frequencies. Different models can be made and compared by claculating and comparing P values for the same sequence. One can plot windows on the X axis and probabilities on the Y axis to visualise results.

r_ab = P(x_i = b | x_i - 1 = a) = n_ab/n_a

n - 1

n = number of replicates

A measure of uncertainty and randomness. If you are certain what the next nucleotide will be, entropy is low. Identified with the Shannon entropy score. Maximized when the probability of all nucleotides is equal; the maximum value is 2. A bit is the log base of 2 of a number. Highly conserved sequences have low entropy

H(X) = - Σ_i p_xi log₂ p_xi

X = vector string of nucleotides

i = index for each site

pxi = probability of nucleotide x_i.

An output file of Cuffdiff.

1. Isoforms.fpkm_tracking. Transcript FPKMs.

2. Genes.fpkm_tracking. Gene FPKMs, tracks the summed FPKM transcripts sharing a gene_id.

3. Cds.fpk_tracking. Coding sequence FPKM, tracks the summed FPKM of transcripts sharing a p_id.

4. Tss_groups.fpkm_tracking. Primary transcript FPKM, tracks the summed FPKM of transcripts sharing a tss.id.

A measure of RNA abundance, controlling for length of exons in the sequence. Reported by Cufflinks. Used as a measure of gene expression. Attempts to normalize the number of reads mapped to a gene by the length of the gene and the total number of mapped library reads. Measures fragments; paired end reads count for two, and single end reads count for one. The larger the difference of FPKM, between two genes, the more confident that one gene is expressed than the other.

FPKM = 10⁹ (C / NL)

C = number of read pairs from transcript

N = total number of mapped read pairs in the library

L = number of exon bases for transcript.

The frequency of heterozygotes expected if two populations are randomly mating, compared to the frequency of heterozygotes within each separate population. Maximum value is 1, indicating no shared alleles at all; the populations are completely different. Minimum value is 0, indicating no allelic differences; the populations are exactly the same. A value of 0.2 or greater is considered a meaningful amount of difference.

F_ST = (H_T - H_S) / H_T

A bioinformatics pipeline.

1. Extract DNA

2. Digest with restriction enzymes

3. Sequence

4. Analyse

The redution in entropy after some information is received. Represented by the total height of bases in a MEME output. First measured in communication cables that went across the ocean; you get a lot of random noise through the cables. The reduction in entropy once a signal has been received.

I(X) = H_before - H_after

"Microarray analysis shows that some microRNAs downregulate large numbers of target mRNAs"

Research goals were to determine if single miNRA can reduce transcript level of multiple mRNA molecules in a tissue specific fashion. Injected HeLa cells with miRNA and monitored gene expression, estimating mRNA abundance with a microarray Non-transfected HeLa cells were used as a control. Are the genes that are downregulated representative of a certain tissue? Is gene downregulation due to direct miRNA binding to mRNAs? Focused on two human miRNAs: miRNA-1 and miRNA-124.  Transcript levels were significant, p < 0.001. Concluded that miRNAs do affect transcript abundance. Obtained a matrix of gene transcript abundance for all genes, from 46 tissues, and ranked level of expression. Funded by a pharmaceutical company in hopes of developing an miRNA drug that can silence disease genes. Unfortunately, what they found was not specific enough.

The log 10 ratio of raw signals of green and red fluorescence from each spot on a microarray. A high value is associated with low intensity, and may be spurious if the numbers are small. Used to determine if treatments have an affect. For each spot, the centre of the M ratio is given by its mean, and variability by its variance. If variance is high, the mean is not a good metric.

M = log₁₀ (RED / GREEN) = log₁₀ RED - log₁₀ GREEN = log (Cy5 / Cy3)

Mean (x bar) = (Σ x_i) / n

Variance (x) = s² = (Σ (x - x bar)²) / (n - 1)

Plot M versus A.

A = (log₁₀ Cy5 + log₁₀ Cy3) / 2 = log₁₀ √(Cy5 x Cy3)

Calculates the transition probabilities between different sequence residues. Can be used to evaluate the probability of a sequence, given a model of the sequence evolution.

P(X) = q_xiΠ^N_{(i = 2)}r_{x(i - 1)xi}

X = sequence of amino acids, nucleotides, or states

q_xi = probability of the first amino acid, nucleotide, or state

r_{x(i - 1)xi} = transition probability between two adjacent amino acids, nucleotides, or states.

N = number of amino acids, nucleotides, or states in X.

A representation of something. Includes qualitative, quantitative, and probabilistic models. Output depends on the equation and parameters used. A simple model assumes that sequence sites are independent, and sequences occur at equal frequency.

y = ax₁ + b

y = ax₁ + bx₂ + c

y = response variable

x = predictive variable

a, b, and c = parameters

A file produced by TopHAt

Exit status = 0. This is good.

Elapsed time = 5.5 hours.

Memory used = 8.6 G.

A file produced by TopHat

Accepted_hits.bam. A list of read alignments. Needs to be converted to .sam to visualise what is here.

Align_summary.txt

Deletions-bed and insertions-bed. Lists the insertions and deletions reported by TopHAt.

Junctions.bed. Tracks the junctions reported by TopHat.

Prep_reads.info

Unmapped.bam

Logs (directory) -run.log, tophat.log, et cetera

A model which assumes that residues are independent, so the probability of a series of events is the produt of probability of each event. Useful for comparing different models that have different parameter estimates. Includes A and T rich repeated sequence model and random sequence model.

P(x | M) = Πⁿ_{(i = 1)}q_xi

x = the sequence

| = "given"

M = model that is assumed

n = length of sequence

i = pointer for the residue in the sequence

q_xi = the probability of that residue

Quantitative insights into microbial ecology

Pronounced "chime"

A free open-source software created on a Python platform. Able to demultiplex large amounts of microbial data generated by high-throughput sequencing technologies such as 454 and Illumina. Analyzes sequences from ribosomal 18S (for eukaryotes) and 16S (for prokaryotes) against databases of reference genomes in order to sort species contained within a sample to operational taxonomic units (OTUs). Allows for contrast of alpha and beta diversity. Python script is easy but limited. Has a very specialized purpose, and is good for only certain types of research.

A mathematical description of a phenomenon.

Response = f(inputs)

Model 1

Assuming all nucleotides normally occur at equal frequencies. The probability of A is 0.25, G is 0.25, T is 0.25, and C is 0.25.

P ("AGCTAT" | M) = (0.25)⁶ = 0.000244.

Used to assay mRNA abundance within a sample. The current technology. A population of RNA fragments are converted into DNA with adapters on the ends, and then sequenced. There are 10s of millions of cDNAs, all attached to a single slide. Produces single and paired end reads. Around 10 - 20 million reads are made for different sequences. Abundance of mRNAs is estimated by number of reads. Introns are not sequenced. Some reads span exons, and need special alignment software such as TopHat, and then Cuffdiff.

1. Align everything that aligns well.

2. For remaining reads, see if they map if split over an interlocus. Parameters include anchor length, anchor mismatch, minimum intron length, and maximum intron length. SAM file output. Criteria attempt to minimize false negatives and false positives.

3. Assemble or summarize information to get an estimate of transcript abundance.

Can compare two models. Comparing the models for A and T rich vs random sequences, we can score A or T as 0.4/0.25 = 1.6, and C or G as 0.1/0.24 = 0.3. For "AGCTAT", S = (1.6)⁴(0.4)² = 1.049.

S = (P(x | M)) / (P(x | M)) = Σ^N_{(i = 1)}log(p_xi / q_xi)

Defines entropy for a string of nucleotides or amino acids. Probability of a nucleotide is estimated by counting the number of times a certain nucleotide occurs within the alignment, and dividing the number by the total number of nucleotides. The value is expressed in log₂; multiply the natural log of p by 1.4427.

H(X) = - Σ_i p_xi log (p_xi)

p_xi = cxi / (Σ_{(i = {A, G, C, T})}c_xi

i = an index for each possible nucleotide

x_i = refers to that nucleotide of i

p_xi = the probability of that nucleotide

c = the number of counts

Spliced transcripts alignment to a referece

An RNA sequence alignment software. Effective when the transcript has introns, mismatches, and indels. High speed, sensitivity, and precision. Can read complex RNA arrangements. Low computational overhead, but uses more RAM and memory.

1. Seed searching for maximum mappable prefix

2. Clustering, stitching, and scoring

A standard method with which to calculate significant changes in microarrays. The t statistic indicates the importance of a mean depends on the spread of values used to compute it. Given that the null hypothesis is true, t values will fall within a range defined by the t distribution and degrees of freedom. Approximately normally distributed. The higher the number, the more confidence there is a difference between treatments.

t = (xbar (√n)) / s

T = (E (log Y)) / (var (log Y))

Chromatin Interaction Analysis by Paired-End Tag

A method of studying chromatin interactions genome-wide. Maps interacting regions with high resolutions, where they occur and at what frequency. Helps understand regulation of gene expression.