Bioinformatics

Requirement analysis

system requirements, database requirements, and system design

Data modeling

conceptual data model, logical data model, and physical data model

Database construction

schema implementation, data population, and query optimization

Application development

database interfaces, data mining tools, and system testing

**Metabolome**

refers to the sum total of all the low-molecular-weight metabolites produced by an organism, cell or tissue

HTTP: HyperText Transfer Protocol

an application-layer protocol for transmitting hypermedia documents, such as HTML. It was designed for communication between web browsers and web servers

SQL

a domain-specific language used in programming and designed for managing data held in a relational database

DBMS

Database Management Systems (DBMS) are software systems used to store, retrieve, and run queries on data

Coding continuity

protein-coding sequences (exons) can be interrupted by noncoding introns, which are removed by splicing from the primary RNA transcript

Alternative splicing

allows for various combinations of exons to be joined to form different mRNAs, which produce more than one polypeptide from a gene

Besides repetitive DNA, there are also a large number of single-copy noncoding regions

* **Gene regulatory sequences**:  e.g., promoters and enhancers
* **Pseudogenes**:  dysfunctional gene copies with significant mutations, usually not transcribed
* **Noncoding RNA genes**:  e.g., rRNAs, tRNAs, microRNAs, long noncoding RNAs (lncRNAs), etc

Repetitive Sequences

* comprise >50% of the genome
* Interspersed repeats are repetitive sequences that are scattered around the genome
* Retrotransposons amply via RNA intermediates
* LINEs (long interspersed nuclear elements, > 5 kb)
* SINEs (short interspersed nuclear elements, < 500 bp) LINEs and SINEs are dispersed throughout the genome, and constitute

Analysis of Repetitive DNA

* RepeatMasker has been the most widely used tool for characterizing repetitive DNA
* Includes Repbase (prototypic)
* Dfam (HMM transposable elements)

Intrinsic gene prediction (ab initio)

methods search for exons and introns based on signals and patterns in the genomic DNA

Extrinsic methods

utilize external data such as transcriptional data and protein sequences like using homogly and comparative based prediction,

Combiners

Intrinsic and extrinsic methods are
used in combination

Framework for Ab Initio Gene Prediction

* The initial (5′) exon is preceded by a core promoter with sequence elements such as the TATA box
* Internal exons are free of in-frame stop codons, and are delimited by splice signals such as 5′ GT and 3′ AG
* The final (3′) exon often contains a stop codon, followed by a polyadenylation signal

Hidden Markov Models (HMMs)

A class of probabilistic models that are generally \n applicable to time series or linear sequences, used in many areas of bioinformatics, including gene prediction and protein domain modeling

GENSCAN

an HMM-based program using many higher-order properties of genomic sequences such as gene density, exon size distribution, etc.

FGENESH

HMM-based gene structure prediction
(multiple genes, both chains)

AUGUSTUS

a generalized HMM with a new method for modeling intron length distributions, used for ab initio gene prediction, and has a flexible mechanism for incorporating extrinsic information, such as EST and protein alignments (most accurate)

JIGSAW

an integrative gene prediction program \n that combines the outputs from other gene finders, \n splice site predictors and sequence alignments

Pairwise Sequence Alignment

Sequence alignment is the identification of residue-residue correspondence

Dotplot

a simple picture that gives an overview of the similarities between two sequences but does not provide a robust statistical measure for the alignment quality. It can be used to find regions of local alignment, insertions, or deletions

Measures of Sequence Similarity For Nucleotide Seqeunces

\+1 for a match, and -1 for a mismatch

Measures of Sequence Similarity For Protein Seqeunces

* PAM (Point Accepted Mutation)
* BLOSUM (BLOcks SUbstitution Matrix)

PAM Substitution Matrices

* Measure of sequence divergence
* 1 PAM = 1% accepted mutation (\~99% sequence identity)
* Higher number means less similar
* The PAM250 level (\~20% overall identity) is the lowest sequence similarity for which a correct alignment may be produced by pairwise sequence comparison alone

BLOSUM Substitution Matrices

* Based on substitution patterns in conserved motifs
from the BLOCKS database
* To avoid overweighting closely related sequences,
groups of proteins with sequence identities higher
than a threshold are replaced by either a single
representative or a weighted average
* The threshold 62% produces the commonly used
BLOSUM62 substitution matrix
* Higher means more similar
* Lower means less similar
* BLOSUM substitution matrices have superseded
the PAM matrices

Gap opening penalty (G)

is the cost of creating a gap

Gap extension penalty

is the cost of extending the gap by one position

\n Affine gap penalty

is calculated as (G + Ln), where n is the length of the gap

Global alignment algorithms

compare two sequences along their entire length, and are most applicable to highly similar sequences

* Based on the mathematical technique called dynamic programming (The Needleman-Wunsch algorithm)
* the time required to align two sequences of lengths, m and n is proportional to (M x N)

Local alignment algorithms

find the most similar regions in two sequences, and are best for sequences that share some degree of similarity or for sequences of different lengths (fast)

* The Smith-Waterman algorithm is the most rigorous method by which regions of two protein or DNA sequences can be aligned
* BLAST

Basic Local Alignment Search Tool (BLAST)

a rapid, heuristic version of the Smith-Waterman \n algorithm

* BLAST takes a small integer w (“word size”), and determine all instances of the words in a query sequence that occur in any sequence in the database

P value

is the probability that the observed match could have happened by chance

E value

an alignment is the expected number of matches that give the same Z score or better if the database is searched with a random sequence

* low E-Values are good

BLAST P

compare a protein query to a database of proteins

protein →(1) protein

BLASTN

compares both strands of a DNA query against a DNA database

DNA →(1) DNA

BLASTX

translates a DNA seqeunce into six protein sequences using all six possible reading frames, and then compares each of these proteins to a protein database

DNA → (6) Proteins

TBLASTN

translate every DNA sequence in a database into six potential proteins, and then compare your protein query against each of those translated proteins

protein →(6) DNA

Too many results in Blast

* Select a RefSeq database to reduce redundancy
* Limit the results by an organism or group
* Adjust the scoring matrix
* Use just a portion of the query sequence

Too few results in Blast

* Raise the expect value threshold
* Try scoring matrices with lower BLOSUM numbers or higher PAM numbers
* Search additional databases
* Use specialized BLAST programs

PSI-BLAST

Position-Specific Iterated BLAST (PSI-BLAST) is particularly well suited for identifying distantly related proteins, which may not be found using BLASTP where hits, along with the query sequence, are used to \n construct a Position-Specific Scoring Matrix (PSSM) in \n an automated fashion

* The PSSM then serves as the query for re-searching the target database
* default is 0.005 for E-value cut off at NCBI

Multiple Sequence Alignment (MSA)

a collection of three or more protein (or nucleic acid) sequences that are partially or completely aligned

MSA can reveal sequence conservation patterns

* Homologous residues may be aligned in columns across the length of the sequences


* Amino acid residues of different physiochemical types are often displayed in different colors

Exact methods (MSA)

employing dynamic programming, not feasible in time and space for many sequences

Progressive alignment (MSA)

ClustalW has been the most popular program from the 1990s until recently, entails calculating pairwise alignment
scores between all the protein sequences being
aligned, and then beginning the alignment with the
two closest sequences and progressively adding
more sequences to the alignment

* cannot go back and change the past alignment
* the final alignment depends on the order in which sequences are joined, and it is therefore not guaranteed to provide the most accurate alignment

Iterative methods

* Clustal Omega, MAFFT and MUSCLE are fast and highly accurate
* compute a suboptimal solution using a progressive alignment strategy, and then modify the alignment using dynamic programming or other methods until a solution converges

Consistency-based methods

\
* ProbCons andT-COFFEE are slower but more accurate
* The idea of consistency is that for sequences x, y
and z, if residue xi aligns with yj and yj aligns with
zk, then xi should align with zk
* score pairwise alignments in the context of information about multiple sequences

Structure-based methods

* using protein structural information to improve the alignment accuracy
* Tertiary structures evolve more slowly than
primary sequences
* It is possible to improve the accuracy of multiple sequence alignments by including the three- dimensional structure information of proteins
* Structure-based methods include PRALINE and T -COFFEE Expresso

Protein Domains

* Domains represent evolutionarily conserved amino acid sequences carrying functional and structural information
* Protein domain analysis helps understand the biological function of a gene
* HMMs are used for protein domain modeling

Pfam

a database of multiple sequence alignments and hidden Markov models (HMMs) of common protein domains

HMMER for Building and Searching HM

Source code and binary distributions of the HMMER \n software package

GenBank

* Nucleotide Sequences
* contains an annotated collection of all publicly
available DNA sequences

UniProtKB

* Protein sequences
* a comprehensive
resource for protein sequence and functional
information, and is participated by:
* Swiss-Prot
* manually annotated
* TrEMBL
* automatically annotated
* Proteomics

GEO

Gene expression profiles

PDB

Protein and nucleic acid structures

KEGG

Biological networks and pathways

PubMed

Publication

RefSeq

aim to provide a high-quality (curated by experts), comprehensive, non-redundant set of sequences

* NG, NM, NP, NR, XM, XP

Database federation

Integrates different data sources using an application-specific, generalized schema, but leaves data at the remote sources (NCBI Entrez)

* Advantages:
* current data
* Flexible architecture
* No data Consolidation
* Disadvantages
* slower queries
* Complex schema
* Data at the sources
* \

Data Warehousing

Consolidates all specified data into a local database with a generalized schema (UniProt, UCSC, Flybase)

* Advantages
* Clean/curated data
* Fast queries
* Data in a local DB
* Disadvantages
* stale data
* complex schema
* data consolidation

Entez Search Statement

\n “search term” \[tag\] Boolean-operator “search term” \[tag\] .

Boolean Operator

And, Or, Not