Altklausuren

• common format for data exchange between different tools and an extension to the FASTA format

• can additionally store a numeric quality score of nucleotides besides other informations about the sequence

The FASTQ file has normally 4 lines per sequence:

• Line 1: “@” + sequence identifier + optional description (like a FASTA title line)

• Line 2: the raw sequence letters

• Line 3: begins with a “+” and is optionally followed by the same sequence identifier (and any description)

• Line 4 encodes the quality values for the sequence in Line 2, and must contain the same number of symbols as letters in the sequence

GeneMark is a gene prediction tool for prokaryotes. When a gene is transcribed, the sense or anti-sense strand is used as a template. Due to the fact that genes can therefore be located on the sense or anit-sense strand, GeneMark includes the information on which strand + or - the gene is predicted.

“+” for forward strand (5’ to 3’)

“-” for reverse strand(3’ to 5’)

Sequence length normalization in next-generation sequencing (NGS) refers to adjusting read counts based on the length of the sequence to allow for accurate comparison across samples.

Purpose:

• correct for bias

• ensure longer sequences do not disproportionately affect the analysis

Common Metrics:

• RPKM (Reads Per Kilobase of transcript per Million mapped reads)

• FPKM (Fragements Per Kilobase of transcript per Million mapped reads)

• TPM (Transcripts Per Million)

• cassette exon (exon skipping): an exon may be spliced out of the primary transcript or retained

• mutually exclusive exons: one of two exons is retained in mRNAs after splicing, but not both (are “competing”)

• competing 5’ splice site (alternative donor site): An alternative 5’ splice junction is used, changing the 3’ boundary of the upstream exon

• competing 3’ splice site (alternative acceptor site): An alternative 3’ splice junction is used, changing the 5’ boundary of the downstream exon

• retained intron: A sequence may be spliced out as an intron or simply retained. This is distinguished from exon skipping because the retained sequence is not flanked by introns.

• multiple promoters

• mutiple poly(A) sites

Two possible approaches:

1. Energy minimization methods: choose complementaey sequence sets that provide the most energetically stable molecules

2. take into account patterns of base-pairing that are conserved during evolution

Energy minimization method:

• every base is compared for complementarity to every other base

• The energy of each predicted structure is estimated by the nearest-neighbor rule:

  • sum the negative base-stacking energies for each pair of bases in the predicted double-stranded regions

  • add positive energies of destabilizing (unpaired) regions

• The complementary regions are evaluated by a dynamic programming algorithm to predict the most energetically stable molecule

• Programs: e.g. MFOLD, ViennaR

N50: The shortest contig length that needs to be included for covering 50% of the genome.

• high value = more continuous assembly with longer contigs, shorter gaps, overall more complete assembly

L50: The count of smallest number of contigs whose length sum makes up half of genome size.

• lower value = more continuous assembly, with fewer contigs needed to reach the 50%, more efficient and better quality assembly

—> to measure the quality of genome assembly

The Ka/Ks ratio is the ratio between the nonsynonymous rate of substitution (Ka) and the synonymous rate of substitution (Ks), to test for natural selection / evolutionary tendencies on genes or proteins.

The Ka/Ks ratio is important to predict if a mutation or rather a change is going to be fixed (evolutionary advantage) or being lost (negative selection).

The Ka/Ks ratio can indicate the following tendencies:

• Ka/Ks > 1 : positive selection

• Ka/Ks = 1 : neutral selection

• Ka/Ks < 1 : negative selection -> purifying selection

• for pseudogenes Ka/Ks = 1 expected

• experimental < 1: underestimated Ka/Ks as genes were compared with present day genes and not the ancestral functional gene that gave rise to the processed pseudogene

• The similarity-based approach uses known genes in one genome to predict (unknown) genes in another genome by e.g. comparison of genomic sequence with homologous genomic sequence from close organisms.

• Given a known gene (or a protein) and a genome sequence, find a set of substrings of the genomic sequence whose concatenation best fits the gene.

• e.g. the known frog gene is aligned to different locations in the human genome. Find the “best” path to reveal the exon structure of human gene. Chaining these local alignments and look for a maximum chain of substrings.

—> possible due to evolutionary conservation of gene sequences across different species

• Example programs: TWINSCAN, SGP2

• Often alignment tools like BLAST or BLAT can be used.

• Sequences are scanned for overrepresented string of certain length

• Challenge: to determine optimal size of an oligo (k-mer) and the number of mismatches allowed

1. Generate K-mers: Slide a window of length k across the genome sequence to generate all possible k-mers.

2. Count Occurrences: Count the frequency of each k-mer in the genome.

3. Identify Repeats: K-mers that appear more frequently than expected are identified as potential repeats.

4. Analysing the repeats by finding out their position in the genome and mapping them

   for a better understanding of the distribution and context of the repeats + possibly further analyses to differentiate the different types of repeats (tandem/dispersed repeats etc.)

Applications:

• Detecting Repeats: Helps in identifying repetitive sequences in the genome, such as tandem repeats, interspersed repeats, and low-complexity regions.

• Genome Assembly: Assists in resolving ambiguous regions during genome assembly by highlighting repetitive sequences.

• we don’t know the motif sequence

• we don’t know the location relative to gene start

• we don’t know the length of the motif

• Motifs can differ slightly from one gene to the next / can have mutations (how many mutations are allowed?)

• How to discern from „random“ motifs?

(Consensus sequences help in finding motifs)

—> These and other problems mean that the computing algorithms have a high runtime and data storage capacity

Definition: The process of identifying (e.g. using RepeatMasker) and masking repetitive sequences in a genome to prevent them from interfering with genomic analyses.

Purpose:

• Improve Assembly: Helps in accurate genome assembly by reducing ambiguity caused by repeats.

• Enhance Annotation: Improves gene prediction and annotation by distinguishing between coding and non-coding regions.

• Data Quality: Ensures high-quality sequence alignment and variant calling by excluding repetitive regions that can cause errors.

1. Conservation: Substitutions in highly conserved regions are more likely to affect protein function, indicating these regions are crucial for the protein's structure or function.

2. Physicochemical Properties: Substitutions that drastically change the physicochemical properties (such as charge, size, or hydrophobicity) of the amino acid are more likely to impact the protein's function or stability.

1. Read Generation: Sequencing the genome to produce short DNA sequences (reads) using technologies like Illumina or PacBio.

2. Read Quality Control: Filtering and cleaning the reads to remove low-quality sequences and contaminants.

3. Read Overlapping: Finding overlaps between reads to identify how they connect, e.g. using De-Bruijn Graph or Overlap-Layout-Consensus (De novo or reference-based)

4. Contig Assembly: Merging overlapping reads into longer contiguous sequences (contigs).

5. Scaffolding: Ordering and orienting contigs into larger structures (scaffolds) using paired-end reads or long-read technologies.

6. Gap Filling: Closing gaps within scaffolds to produce a more complete genome sequence.

7. Quality Assessment: Evaluating the assembly for accuracy, completeness, and contiguity using metrics like N50 and L50.

8. Optional annotation by identification of genes (MAKER, AUGUSTUS) and functional annotation (BLAST)

TargetScan:

• thermodynamics-based modeling of RNA:RNA duplex interactions

• comparative sequence analysis

Input:

• miRNA that is conserved in multiple organisms

• a set of orthologous 3’ UTR sequences from these organisms

Steps:

1. search the UTRs in the first organisms for segments of perfect Watson-Crick complementarity to bases 2-8 of the miRNA: “miRNA seed” and “seed matches”

2. extend each seed match with additional base pairs to the miRNA as far as possible in each direction, allowing G:U pairs, but stopping at mismatches

3. optimize basepairing of the remaining 3’ portion of the miRNA to the 35 bases of the UTR immediately 5’ of each seed match using the RNAfold program

4. assign a folding free energy to each such miRNA:target site interation

5. assign a Z score to each UTR

6. sort the UTRs in this organism by Z score and assign a rank to each predict

7. predict as targets those genes for which both Zi >= Zc and Ri <= Rc for an orthologous UTR sequence in each organism, where Zc and Rc are pre-chosen Z score and rank

A pan-genome is the entire set of genes from all strains within a species or clade. It includes:

• Core Genome: Genes shared by all strains.

• Accessory Genome: Genes not present in all strains, including:

  • Dispensable Genes: Present in some but not all strains.

  • Unique Genes: Specific to individual strains.

Purpose: To study genetic diversity, evolution, and functional adaptations.

• Bakterien haben also eine höhere Gendichte als höhere Eukaryoten

• Gendichte bei Prokaryoten (Bakterien & Archaea): ~90% >>> Gendichte bei Eukaryoten: ~1-2%

• In Bakterien liegt die Gendichte bei 1 Gen pro 1.000 - 1.4000 Basen

• In höheren Eukaryoten liegt die Gendichte bei 1 Gen pro 100.000 Basen

• EM Algorithmen, wie z.B. MEME (Multiple EM for Motif Elucidation), starten bei einer Site von mehreren Sites von Zielsequenzen und wechseln sich dann ab zwischen der Zuordnung der Site zu einem Motiv und dem Updaten des Motivmodells. Dabei werden nur die besten Treffer pro Sequenz angezeigt, obwohl niedrigere Treffer in der gleichen Sequenz auch einen Effekt haben können.

• start with initial guesses for region and size (e.g. region of a binding size is already known from prior experiments)

1. expectation step:

   • position-wise composition of the site is used to estimate the probability of finding the site at any position of the seqs

   • these probabilities are used in turn to provide new information as to the expected base distribution for each column

1. maximization step: new counts of bases for each position in the site found in E-step are substituted for the previous set

• E- and M-steps repeated until convergence (no more changes)

• Result:

  • best location of the size in each seq

  • best estimate of the base composition of each column in the site

• sequence logo height is showing the frequencies scaled relative to the information content (measure of conservation) of the base at the position

• can be corrected by base frequencies of the bases

• data might include pseudocounts to overcome effects of missing data

• the maximum value for DNA bases is 2 bits (log2(4)) —> perfectly conserved

• pseudogenes: Nonfunctional sequences of genomic DNA that are originally derived from functional genes (by gene duplication), but exhibit such degenerative features as premature stop codons and frameshift mutations that prevent their expression

• Pseudogene sind dennoch wichtig, da einige eine Rolle bei der Regulierung der Genakativität spielen und somit nicht funktionslos sind.

• might interfer with experiments

  • PCR and hybridization experiments

  • transcribed pseudogenes

  • interference with disease diagnostics and treatment

• molecular record of dynamics and evolution of genomes

  • rate of nucleotide substitutions

  • rate of DNA loss

• improvement of gene prediction and annotation efforts

• multiplicity: one miRNA can target more than one gene

  • some miRNAs appear to be very promiscuous, with hundreds of predicted targets, but most miRNAs control only a few genes

• co-operativity: one gene can be controlled by more than one miRNA

  • Some target genes appear to be subject to highly cooperative control, but most genes do not have more than four targets sites

Positive Predictive Value (PPV) = Maß dafür, wie wahrscheinlich es ist, dass ein positiv vorhergesagtes Ereignis tatsächlich eintritt.

PPV = TP / TP + FP

Wenn das Target stark mit dem Informanten übereinstimmt, erhöht sich der PPV. Dies liegt daran, dass die Anzahl der korrekt vorhergesagten positiven Fälle (TP) steigt und die Anzahl der falsch positiven Vorhersagen (FP) sinkt.

• Durch Verwendung einer anderen Stelle für die Translationsinitiation (alternative Initiation)

• alternatives Exon hat Stopp-Codon (alternative Terminierung)

• Eine andere Translationsterminationsstelle aufgrund eines Frameshift (Verkürzung oder Verlängerung)

• Ändern des inneren Bereichs aufgrund eines in-Frame Insertion oder Deletion

• Operons könnten in termophilen Organismen entstanden sein, da die Organisation von Genen in Operons die Assoziation / Verbindung von funktionell verwandten Protein-Produkten ermöglicht und diese sich somit gegenseitigen Schutz vor thermischen Verfall bieten.

• Rolle des Horizontellen Gentransfers: Vorteil komplette Sets an Genen zu übertragen und dem Empfänger einen definierten Phenotyp zu übertragen

• evtl. ausgehend von thermophilen Bakterien

• Die Genauigkeit der Vorhersage steigt mit der Vergrößerung des Frameshifts der ORF-Länge, da längere ORFs eher den tatsächlichen Genen entsprechen.

• vmtl bezogen auf GeneMark? -> higher sensitivity, lower specificity

• Analysefenster: Größeres Fenster erhöht Sensitivität und False Positives

• Frameshift: Erhöhte Frameshift-Fehler verschlechtern die Vorhersagegenauigkeit

Proteinkomplex: RNA-incduced splicing complex (RISC) enthält das Enzym DICER, welches die pre-miRNA in miRNA spaltet

• Ein starker Promoter ist der Consensus Sequenz sehr ähnlich.

• (ein schwacher Promoter unterscheidet sich stärker von der Consensus Sequenz)

• DNA-Sequenz, die eine hohe Transkriptionsrate ermöglicht

  • die effizient an die RNA-Polymerase bindet und einen robusten Transkriptionsbeginn fördert

  • ein starker Promotor hat eine hohe Affinität für die RNA-Polymerase, was eine effiziente Bindung und Initiierung der Transkription ermöglicht

  • Vorhandensein spezifischer Sequenzmotive innerhalb der Promotorregion

• Sigma Faktor: sind Proteine die Teil des RNA Polymerase Proteinkomplexes sind, welcher an den Promoter bindet.

• Sie werden für die Initiation der Transkription benötigt.

• Häufigster Sigma Faktor: σ^70 (Housekeeping-Sigma-Faktor von E.coli; steuert die Transkription)

• Es gibt mehrere austauschbare Sigma-Faktoren, von denen jeder eine bestimmte Gruppe von Promotoren erkennt (Promotoren von Housekeeping-/Hitzeschock-Genen)

• Clone-by-Clone:

  • physical mapping: requires construction of clone-based physical map

  • assembly: easier to resolve complex genomic regions as position of contigs is already known (due to the physical mapping)

  • labor intensity: physical mapping is labor intensitive, but after mapping clones can be divided between different labs for sequencing

• Whole Genome Shotgun:

  • physical mapping: mapping phase is skipped and subclone library is constructed from entire genome

  • assembly: order / position of contigs needs to be inferred from overlapping reads and read pairs which can be problematic for tandemly repeated DNA (incorrect overlaps)

  • labor intensity: less labor intensive, but requires more computational resources

Clone-by-clone shotgun: BAC clones werden benötigt, clones vorher mappen

Whole genome shotgun: Mapping-Phase wird übersprungen, Assembly dauert länger

1. Auswahl eines Klons (z.B. BAC).

2. Reinigung und physikalische Fragmentierung der BAC-DNA.

3. Subklonierung der DNA-Fragmente (2-5 kb).

4. Erstellung von Sequenz-Reads aus Subklonen (mehrere tausend Reads pro BAC).

5. Assemblierung der Reads basierend auf Sequenzüberlappungen zu einer vorläufigen Sequenz.

6. Identifikation und Ausbesserung von Lücken und Bereichen mit schlechter Sequenzqualität durch zusätzliche Sequenzdaten.

• Die mapping Phase wird übersprungen

• Shotgun Sequenzierung wird fortgesetzt unter Verwendung von Subklon-Bibliotheken die aus dem gesamten Genom hergestellt werden

• Typischerweise werden zig Millionen von Sequenz-Reads erstellt

• Computergestützte Anwendungen werden verwendet, um Contigs aus den verschiedenen Reads zu erzeugen.

• Entstehende Lücken werden durch anschließende Verfahren geschlossen, um eine vollständige genomische Sequenz zu erhalten.

Whole Genome Shotgun (WGS) für prokaryotische Genome:

• Prokaryotische Genome sind kleiner und weniger komplex.

• WGS ist schneller und kostengünstiger, da die Kartierungsphase entfällt.

• Weniger repetitive DNA-Sequenzen erleichtern die Assemblierung.

Clone-by-Clone für eukaryotische Genome:

• Eukaryotische Genome sind größer und komplexer.

• Physische Kartierung hilft bei der Bewältigung der Komplexität und erleichtert die Assemblierung.

• Handhabung repetitiver Sequenzen durch physische Kartierung.

• Effiziente Arbeitsaufteilung durch Verteilung der Klone auf verschiedene Labore.

• Erleichterte Assemblierung komplexer genomischer Regionen.

(Approaches can be combined in a hybrid shotgun-sequencing approach)

Genomweite Analyse:

1. Genomsequenz-Assemblies und EST-Sequenzen

2. EST Clustering von UNIGENE

3. BLAST-Suche:

   • Kandidaten-Gen-Regionen (BLAST threshold < E10-5)

   • Vermeintliche (kurze) Exons (BLAST threshold < E10-10)

4. Alignment der genomischen Regionen mit ESTs durch dynamische Programmierung

5. Splicing-Erkennung durch computergestütztes Verfahren

Verifizierung:

1. RNA-Isoformen-Analyse mittels RT-PCR (unterschiedliche PCR-Produktlängen)

2. Mikroarrays mit Exon-Exon-Junction-Probes

Consequences of new protein parts:

• alter protein binding properties, e.g. receptor / ligand

• alter intracellular localization, e.g. membrane insertion

• alter extracellular localization, e.g. secretion

• alter enzymatic or signaling activities

• alter protein stability, e.g. inclusion of cleavage sites

• Insertion of post-translation modification domains

• Change ion channel properties

• Influence RNA function

  • AS does occur to alter 5’ and 3’ UTR regions - Proposed roles in subcellular localization and RNA stability

• Coordinated Regulation of Biological Events

  • Neuron development (Dscam)

  • Channel activity associated with hearing (slo)

  • Muscle contraction

  • Neurite growth

  • Cell differentiation

  • Apoptosis

1. Intrinsic

   a) Open reading frames (ORFs)

   b) Codon usage

   c) Anwesenheit von RBS (ribosomal binding sites)

   d) Periodizität von repeats (Wiederholungen)

2. extrinsic

   a) Expressed Sequence Tags (ESTs)

   b) cDNA-Alignments

   c) homology in known Exons

GenScan: designed to predict complete gene structures but also partial genes or multiple genes separated by intergenic DNA within a sequence

• based on generalized Hidden Markov Models

• Model both strands at once

GenScan States:

• N – intergenic region

• P – Promoter

• F – 5‘ untranslated region

• T – 3’ untranslated region

• A – poly-A

• E – Exon (sngl = single, init = initial, term = terminal, k = Phase k internal)

• Ik – Phase k Intron: 0 – zwischen Codons, 1 – nach der ersten Base eines Codons, 2 – nach der zweiten Base eines Codons

• Each state may output a string of symbols (according to some probability distribution).

• Explicit intron/exon length modeling

• Special sensors for Cap-site and TATA-box, Advanced splice sites

• uses dynamic programming to determine the most likely gene structure compatible with the given sequence.

Parallel Unsupervised Training and Prediction:

• GenScan initializes all model parameters and uses GeneMark to parse the sequence into "coding" and "non-coding" regions.

• The newly labeled sequences are used to re-estimate the model parameters until the model converges.

Rekursive Definition des besten Scores für eine Subsequenz i, j —> 4 Möglichkeiten:

1. i, j sind ein Basenpaar, hinzugefügt zu einer Struktur für i+1 ... j-1, add +1

2. i ist ungepaart, hinzugefügt zu einer Struktur für i+1...j

3. j ist ungepaart, hinzugefügt zu einer Struktur für i...j-1

4. i, j sind gepaart, aber nicht zu einander: die Struktur von i...j fügt Unterstrukturen zusammen für zwei Untersequenzen, i...k und k+1...j (bifurcation)

• Man könnte die obige Formel noch verbessern, indem man auch Pseudoknots beachtet und mit in die Formel einbaut.

• this base pair maximization will not necessarily lead to the most stable structure

  • additionally use thermodynamic information:

    • negative stacking energy for matches

    • positive destabilizing energies for loops (size-dependend)

  • (minimum free energy method)

• Retrotransposons

  • LTRs (Long Terminal Repeat Retrotransposons)

  • LINEs (Long Interspersed Nuclear Elements) [autonomous]

  • SINEs (Short Interspersed Nuclear Elements) [nonautonomous]

• DNA Transposons

  • TIR (Terminal inverted repeat)

  • MITE (Miniature Inverted-repeat Transposable Elements)

• Tandemly repeated DNA:

  • Microsatellites

  • Minisatellites

  • Cryptically simple repeats

  • Low complexity repeats

  • Satellite repeats

  • Telomeric repeats

• Derived from biologically active “transposable elements” (TEs)

• Involve RNA intermediates (Retroelements) or DNA intermediates (DNA transposons)

• Retroelements: reproduce via reverse transcription followed by integration inot the host DNA (LTR ,LINEs, SINEs)

• DNA transposons: capable of integrating themselves to, and excising themselves from, the host genome, thus taking advantage of the host replication thorugh this “cut-and-paste” mechanism

• 3 different mechanisms for transposition

  • conservative transposition

  • replicative transposition

  • retrotransposition

1. Tandemly repeated DNA (Simple sequence repeats without interuption)

   • Microsatellites

   • Minisatellites

   • Satellite and telomeric repeats

2. Cryptically simple repeats

3. Low complexity repeats

• Interspersed repeats sind mobil und über das Genom verteilt

• anderen repititiven Sequenzen sind stationär und kommen in spezifischen Clustern vor

Interspersed repeats

• Retrotransposons (LINEs, SINEs, LTR)

• DNA-Transposons

• SNP stands for Single Nucleotide Polymorphism, it occurs when a single nucleotide replaces one of the other three nucleotide letters in a genome (or in a DNA sequence).

• SNPs may occur anywhere: Most SNPs are found outside of coding seqs => SNPs found in a coding seq are of great interest as they are more likely to alter function of a protein

• most common type of genetic variation in humans. They account for 90% of the variation between individuals

• Most are neutral polymorphisms, some cause disease

• density = ~1 every 100-300 bases

Coding SNPs: occur within coding region of a gene

• synonymous: not causing a change in the amino acid

• nonsynonymous: alters the amino acid sequence of the protein, potentially affecting protein function (missense or nonsense mutations)

Non-coding SNPs: occur outside the coding regions of a gene

Regulatory SNPs: positions that fall in regulatory regions of genes

Intronic SNPs: positions that fall within introns

SNPs may be informative with respect to disease:

• Functional variation. A SNP associated with a nonsynonymous substitution in a coding region will change the amino acid sequence of a protein

• Regulatory variation. A SNP in a noncoding region can influence gene expression

• Association. SNPs can be used in whole-genome association studies. SNP frequency is compared between affected and control populations.

• Number of miRNA genes present

  • Plants: 100-200 genes

  • Animals:100-500 genes

• Location within genome:

  • Plants: predominantly intergenic regions

  • Animals: intergenic regions, introns

• Presence of miRNA clusters:

  • Plants: uncommon

  • Animals: common

• miRNA biosynthesis:

  • Plants: Dicer-like

  • Animals: Drosha, Dicer

• Mechanism of repression:

  • Plants: mRNA-cleavage (methylation?)

  • Animals: Translational repression

• Location of miRNA-binding motifs

  • Plants: predominantly in the ORF

  • Animals: predominantly in the 3’-UTR

• Number of miRNA-binding sites within target sites:

  • Plants: Generally one

  • Animals: Generally multiple

• Function of known target genes:

  • Plants: Regulatory genes - crucial for development, enzymes

  • Animals: Regulatory genes - crucial for development, structural proteins, enzymes

TargetScan:

• thermodynamics-based modeling of RNA:RNA duplex interactions

• comparative sequence analysis

Input:

• miRNA that is conserved in multiple organisms

• a set of orthologous 3’ UTR sequences from these organisms

Steps:

• search the UTRs in the first organism for segments of perfect Watson-Crick complementarity to bases 2-8 of the miRNA: “miRNA seed” and “seed matches”

• extend each seed match with additional base pairs to the miRNA as far as possible in each direction, allowing G:U pairs, but stopping at mismatches

• optimize basepairing of the remaining 3’ portion of the miRNA to the 35 bases of the UTR immediately 5’ of each seed match using the RNAfold program

• assign a folding free energy G to each such miRNA:target site interaction

• assign a Z score to each UTR

• sort the UTRs in this organism by Z score and assign a rank R to each

• predict as targets those genes for which both Zi >= Zc and Ri <= Rc for an orthologous UTR sequence in each organism, where Zc and Rc are pre-chosen Z score and rank

• Statistical model that captures the patterns of covariation that can be obtained from a MSA. Covariated bases tend to coevolve as this ensures that the base pair is maintained and RNA structure is conserved. RNA structure prediction can be improved by giving positions with greater covariation more weight.

• describes both the secondary structure and the primary sequence consensus of an RNA

• Can be applied ro several RNA anlysis problems:

  • consensus secondary structure prediction

  • multiple sequence alignment

  • database similarity searching

• Iterative training procedure

• Optimal algorithm for RNA secondary structure prediction based on pairwise covariations in multiple alignments

• Covariation ensures ability to base pair is maintained and RNA structure is conserved

• Covariance models are constructed automatically

  • from existing RNA sequence alignments

  • even from initially unaligned example sequences

• Needs to be well trained

• Not suitable for searches of large RNA and for database searches

  • Structural complexity of large RNA cannot be modeled

  • Runtime

  • Memory requirements

• Can be used for scanning candidate RNAs identified by other methods

• sehr rechenintensiv aufgrund der 3D dynamischen Programmierung

• keine Angabe von tRNA-spezifischen Informationen dank allgemeinen Ansatz

Unterschiede:

1. Prokaryotische Genome sind im Allgemeinen wesentlich kleiner als eukaryotische Genome

2. Eukaryotische Genome besitzen einen hohen Anteil an nichtcodierender DNA (etwa 95% im Mensch) wohingegen Prokaryotische Genome nur relativ geringe Anteile nichtcodierender DNA besitzen (ca. 5-20%)

3. Das eukaryotische Genom besitzt eine Intron-Exon-Struktur der Gene, wobei das prokaryotische Genom kaum bis gar keine Introns besitzt

4. Das prokaryotische Genom ist polycistronisch, das eukaryotische Genom monocistronisch

5. Die Gendichte in eukaryotischen Genomen ist niedriger aufgrund der vielen nicht-codierenden Bereiche, in prokaryotischen Genomen ist die Gendichte wesentlich höher

• Size

  • prokaryotes between 1s and 10s of Mb 

  • eukaryotes between 1s and 1.000s of Mb

• Topology: 

  • prokaryotes: mostly circular 

  • eukyryotes: mostly linear 

• Gene number: 

  • prokaryotes: most <10.000

  • eukaryotes: often >10.000

• Pseudogenes: 

  • prokaryotes: few

  • eukaryotes: many

• Complexity:  

  • prokaryotes: low

  • eukaryotes: high

• Horizontal gene transfer:  

  • prokaryotes: frequent

  • eukaryotes: rare

• Intergenic regions:  

  • prokaryotes: short (<100kb)

  • eukaryotes: long (often >100kb)

• Genome duplication:  

  • prokaryotes: none

  • eukaryotes: frequent (especially in plants)

• Gene duplication:  

  • prokaryotes: rare

  • eukaryotes: frequent

• Repeated sequences:  

  • prokaryotes: minor components

  • eukaryotes: major components

Eine Vergrößerung des Windows kann dazu führen, dass mehr potenzielle ORFs erkannt werden

• = erhöhten Sensitivität (True Positives)

• = verringerte Spezifität: erhöhte Anzahl der False Positives , da mehr zufällige Sequenzen als ORFs erkannt werden könnten, die tatsächlich keine Gene sind

=> beeinflusst den positiven Vorhersagewert

Eine optimale Window-Größe muss daher gefunden werden, die das Gleichgewicht zwischen Sensitivität und Spezifität hält, um den höchsten PPV zu erzielen.

• Alternative Initiation der Translation (als ursprüngliche Stelle=

• alternative Terminierung durch Hinzufpgen einer Stopp-Codons in alternativem Exon

• Verkürzung (oder Verlängerung) aufgrund eines Frameshifts

• Ändern des inneren Bereichs aufgrund einer in-Frame Insertion or Deletion

TFFM <-> Transkriptionfactor vorhersage (Transcription Factor Binding Motif Prediction)

Kraken <-> Quality Control (Taxonomic Classification of Metagenomic Sequences)

Prodigal <-> prok. Genvorhersage

Augustus+ <-> euk. Genhorhersage

Annovar <-> Functional Annotation of Genetic Variants (Genomic Variation)

Consequences of new protein parts:

• alter protein binding properties, e.g. receptor / ligand

• alter intracellular localization, e.g. membrane insertion

• alter extracellular localization, e.g. secretion

• alter enzymatic or signaling activities

• alter protein stability, e.g. inclusion of cleavage sites

• Insertion of post-translation modification domains

• Change ion channel properties

Effekt:

• Bildung von Isoformen mit unterschiedlichen Funktionen

• Veränderung der Protein-Interaktionen und Signaltransduktion

• Alignment of ESTs (expressed sequence tags) against DNA (/pre-mRNA?) sequence

• Insertions and deletions in the ESTs relative to the [?pre-] mRNA are identified as potential alternative splices

• Alternative splices are detected when two splices are mutually exclusive

• Requires ESTs which are cDNA sequences derived from mRNA with reverse transcriptase

1. Intrinsic

   a) Open reading frames (ORFs)

   b) Codon usage

   c) Anwesenheit von RBS (ribosomal binding sites)

   d) Periodizität von repeats (Wiederholungen)

2. extrinsic

   a) Expressed Sequence Tags (ESTs)

   b) cDNA-Alignments

   c) homology in known Exons

• DNA motif for protein translation initiation site in most eukaryotic mRNA transcripts

• a region around start codon

• 5’-(gcc)gccRccAUGG-3’

• (eukaryotic equivalent to Shine-Dalgarno)

GeneMarkS: a self-trained method for prediction of gene starts in microbial genomes

GeneMarkS-T: gene prediction in eukaryotic transcripts, additionally uses homology based inference / integrates transcriptome data

—> The number of available transcripts plays a crucial role in the model's accuracy, as more transcripts lead to more and better training data and thus more precise predictions.

• Self-Training method and iterative procedure: algorithm can improve its parameters thorugh iterative anaylsis of the input data

• employs fifth-order Markov Models

• uses Markov chain model to represent the statistics of coding and noncoding reading frames

Struktur, die nicht vorhergesagt werden kann: Pseudoknots

• pseudoknots violate the recursive definition of the optimal score S(i,j)

• Man könnte die obige Formel noch verbessern, indem man auch Pseudoknots beachtet und mit in die Formel einbaut.

• this base pair maximization will not necessarily lead to the most stable structure

  • additionally use thermodynamic information:

    • negative stacking energy for matches

    • positive destabilizing energies for loops (size-dependend)

  • (minimum free energy method)

Location of miRNA-binding motifs within target genes

• Plants: predominantly the ORF

• Animals: Predominantly the 3’ UTR

Number of miRNA-binding sizes within target genes:

• Plants: Generally 1

• Animals: Generally multiple

miRNA Processing

• Plants: Dicer-like 1 enzyme, (Drosha gene that is in animal genomes is absent)

• Animals: Drosha gene + Dicer processes primary miRNA

Site Location

• Plants: Target sites outside the 3’ UTR region of mRNA

• Animals: Target sites typically located within 3’ UTR region of mRNA

TargetScan:

• thermodynamics-based modeling of RNA:RNA duplex interactions

• comparative sequence analysis

Input:

• miRNA that is conserved in multiple organisms

• a set of orthologous 3’ UTR sequences from these organisms

Steps:

• search the UTRs in the first organism for segments of perfect Watson-Crick complementarity to bases 2-8 of the miRNA: “miRNA seed” and “seed matches”

• extend each seed match with additional base pairs to the miRNA as far as possible in each direction, allowing G:U pairs, but stopping at mismatches

• optimize basepairing of the remaining 3’ portion of the miRNA to the 35 bases of the UTR immediately 5’ of each seed match using the RNAfold program

• assign a folding free energy G to each such miRNA:target site interaction

• assign a Z score to each UTR

• sort the UTRs in this organism by Z score and assign a rank R to each

• predict as targets those genes for which both Zi >= Zc and Ri <= Rc for an orthologous UTR sequence in each organism, where Zc and Rc are pre-chosen Z score and rank

Hybridisierungsverfahren (Microarrays):

• Vorteile:

  • Relatively low cost

  • Well established in clinical use

  • relatively fast

• Nachteile:

  • Analysis only of pre-defined sequences

  • Dynamic range limited by scanner

  • high background noise

  • cross-hybridization possible

Sequenzbasierende Verfahren (RNA-seq)

• Vorteile:

  • identification of alternative splice variants / new transcript

  • high sensitivity

  • broad spectrum of applications

• Nachteile:

  • relatively high cost

  • high computational effort

  • prone to contamination

  • complexity of data analysis

Identifies the full set of transcripts, including large and small RNAs, novel transcripts from unannotated genes, rare transcripts, splicing isoforms and gene-fusion transcripts

• Reveals the complex landscape and dynamics of the transcriptome from yeast to human at an unprecedented level of sensitivity and accuracy

• Base-pair-level resolution and a much higher dynamic range of expression levels

Overview of the experimental steps in an RNA sequencing (RNA-seq) protocol

RNA extraction → target enrichment → cDNA → library prep → sequencing → Transcriptome/genome mapping → data analysis 

Experimental design: number of replicates, depth of sequencing

Parameters: alignment rate, desired power, significance level, log-fold change

RNA-seq workflow:

• Quality control

• Alignment of reads to reference genome

• Transcriptome assembly

• Differential expression

ORPHEUS: Gene prediction in bacterial genomes, Genvorhersage

SIFT: Sort intolerant from tolerant substitutions, SNPs

MiRScan: miRNA

Genescan: Genvorhersage von Prokaryoten

REPuter: Repeats

• nicht funktionale Gene aber aus funktionalen Genen hervorgegangen

• besitzen degenerative Eigenschaften (missense / nonsense mutations), die Expression verhindern

• 2 Hauptklassen:

  • conventional

  • processed pseudogenes

• Ka: Zahl der nicht synonymen Mutationen

• Ks: Zahl der synonymen Mutationen

• Ka/Ks höher mit niedrigerer Konservierung

• Ka/Ks = 1 => kein Selektionsdruck

• Ks/Ks > 1 => positiver Selektionsdruck (positive selection)

• Ka/Ks < 1 => negativer Selektionsdruck (purifying selection)

• for pseudogenes Ka/Ks = 1 expected

• experimental < 1: underestimated Ka/Ks as genes were compared with present day genes and not the ancestral functional gene that gave rise to the processed pseudogene

• Content based

  • Beispiel: ORFs, Codon usage, Repeat periodicity, Compositional complexity

• Site based

  • Beispiel: splice sites, TF binding sites, Consensus sequences, Polyadenylation signals, start / stop codons

• Comparative

  • Beispiel: Inference based on homology, Protein sequence similarity, Modular structure of proteins usually precludes finding complete gene

• Ablauf Splicing:

  • 5 critical bases: 5’ splice site / donor splice site (GU), branch point (A), 3’ splice site / acceptor splice site (AG)

  • cleavage on 5’ splice site of pre-mRNA

  • reaction between 5’ splice site and branch site leads to formation of lariat-like intermediate

  • cleavage at 3’ splice site

  • ligation of exons

• Types of AS:

  • constitutive AS: more than one product is always made from transcribed gene

  • regulated AS: different forms are generated under different conditions

• Effekt:

  • Bildung von Isoformen mit unterschiedlichen Funktionen

  • Veränderung der Protein-Interaktionen und Signaltransduktion

What are SNPs?

• SNP stands for Single Nucleotide Polymorphism, it occurs when a single nucleotide replaces one of the other three nucleotide letters in a genome (or in a DNA sequence).

• SNPs may occur anywhere: Most SNPs are found outside of coding seqs => SNPs found in a coding seq are of great interest as they are more likely to alter function of a protein

• most common type of genetic variation in humans. They account for 90% of the variation between individuals

• Most are neutral polymorphisms, some cause disease

• density = ~1 every 100-300 bases

Other types of polymorphisms

• Insertion / Deletions (Indels)

• Copy Number Variation

• Tandem Repeats

• Micro- / Mini-satellites

• Structural Variations

• Transposable Elements

• A sequence of nucleotides in mRNA that is recognized and bound by the ribosome during the initiation of translation

• ca. 3-10 nucleotides before initiation codon

• e.g. Shine-Dalgarno (SD) sequence, which is located 5-10 bp upstream of the start codon (AUG)

• The identification of RBS is crucial for predicting the location of genes in prokaryotic genomes. Since the RBS is closely associated with the start codon of a gene, finding the RBS can help in pinpointing the start site of protein-coding sequences.

• Tools: Glimmer, GeneMark, Prodigal

• RPKM: reads per kilobase of transcript per million mapped reads

• FPKM: fragments per kilobase of transcript per million mapped reads

• TPM: transcrips per million

• Example usage: RNA fragmentation during library construction

Ein Referenzgenom einer Spezies ist das Genom gegen welches neue Sequenzierungsdaten verglichen werden. Es wird als eine Art ideal angesehen und daher wird nach dem Sequenzieren damit verglichen, um bspw. Fehler oder Gaps zu erkennen.

Das Referenzgenom ist daher das “perfekte” Genom für eine Spezies und kann auch modifiziert werden, falls neue Information zu diesem entdeckt werden. Es wird also im Laufe der Zeit potentiell verändert, um up-to-date zu sein.

RNAfold predicts the secondary structure of RNA by finding the structure with the minimum free energy. This structure is considered optimal because it is the most thermodynamically stable configuration.

• Genes in different organisms are similar

• The similarity-based approach uses known genes in one

  genome to predict (unknown) genes in another genome

• Given a known gene (or a protein) and a genome sequence, find a set of substrings of the genomic sequence whose concatenation best fits the gene

• e.g. the known frog gene is aligned to different locations in the human genome —> find “best” path to reveal the exon structure of human gene

• Nonfunctional sequences of genomic DNA that are originally derived from functional genes, but exhibit such degenerative features as premature stop codons and frameshift mutations that prevent their expression

• A fundamental feature of pseudogenes is that their nucleotide sequences differ from those of the paralogous functional genes at crucial points

• Two types of pseudogenes: conventional, processed

Identification Approach:

1. Sequence Comparison: Use BLAST to align sequences with known functional genes and look for mutations.

2. Phylogenetic Analysis: Create phylogenetic trees to compare the evolutionary relationships between functional genes and potential pseudogenes

Beim Splicen werden unter anderem die Introns aus der Sequenz gesplicet / entfernt.

Hierfür gibt es die 3’ splice site und die 5’ splice site. Die 3’ splice site startet das spleißen von vorne der Sequenz wohingegen die 5’ splice site dieses von hinten startet.

• Sequences are scanned for overrepresented string of certain length

• Challenge: to determine optimal size of an oligo (k-mer) and the number of mismatches allowed

1. Generate K-mers: Slide a window of length k across the genome sequence to generate all possible k-mers.

2. Count Occurrences: Count the frequency of each k-mer in the genome.

3. Identify Repeats: K-mers that appear more frequently than expected are identified as potential repeats.

4. Analysing the repeats by finding out their position in the genome and mapping them

   for a better understanding of the distribution and context of the repeats + possibly further analyses to differentiate the different types of repeats (tandem/dispersed repeats etc.)

Applications:

• Detecting Repeats: Helps in identifying repetitive sequences in the genome, such as tandem repeats, interspersed repeats, and low-complexity regions.

• Genome Assembly: Assists in resolving ambiguous regions during genome assembly by highlighting repetitive sequences.

PolyPhen (Polymorphism Phenotyping) predicts the impact of amino acid substitutions on the structure and function of proteins, which helps identify potentially damaging mutations.

Goal: to obtain a lower limit estimate for the quantity of non-synonymous SNPs that might have phenotypic effects

1. Map known disease mutations onto known 3D structures of proteins

2. Compare results with a control set of substitutions observed between these proteins and their closely related homologs from other species that are unlikely to cause severe effects on the phenotype

3. Map a large number of non-synonymous SNPs onto protein structures: thought to be neutral or to be the cause of only minor phenotypic effects

• we don’t know the motif sequence

• we don’t know the location relative to gene start

• we don’t know the length of the motif

• Motifs can differ slightly from one gene to the next / can have mutations (how many mutations are allowed?)

• How to discern from „random“ motifs?

(Consensus sequences help in finding motifs)

—> These and other problems mean that the computing algorithms have a high runtime and data storage capacity

• Produces a permutation of a string that is easier to compress

• it is reversible: the original string can be recovered

• Applications:

  • string compression

  • pattern matching

  • searching for patterns in strings

• Approach:

  • form successive circular permutations of the string

  • sort these lines into alphabetical order

  • Report the last column

• The BWT brings repeats together, facilitating compression

—> Example tool that uses BWT is Bowtie

MirScan:

• a bioinformatics tool for predicting miRNA target sites in mRNA sequences by analyzing sequence complementarity and conservation across species

• Finds potential miRNA target sites by aligning miRNAs to target mRNA sequences.

Genomics: the study of the complete set of DNA sequences (the genome) in an organism, including all of its genes and non-coding regions

• identify DNA sequences of an organism’s genome

• Understanding genetic variations

• identify genes associated with dieases

• example tools: BLAST, Genome Browser

Proteomics: the large-scale study of proteins, particularly their functions, structures, and interactions within a cell or organism.

• measuring protein abundance and expression levels

• exploring protein functions, interactions, modifications

• example tools: STRING, Mascot

Transcriptomics: the study of the complete set of RNA transcripts produced by the genome under specific circumstances or in a specific cell type.

• analyzing gene expression and understanding how genes are regulated.

• example tools: STAR, DESeq2