Lecture & Exercises

• 70% transcribed

• 2% coding

• Yes

• linear correlation

• Human 11 genes per MB (10⁶ bases) / Fly 76 / Yeast 496

• 44%

• Clone by clone:

  • easy assembly

  • more work

• Whole genome shotgun:

  • complicated assembly (tandemly repeated DNA, Genomewide repeats)

  • littel preparation needed

• No

• ribosome binding sites

• TF binding sites

• nuclease binding sites

• Prokaryotes have few TFs

• Prokaryotes have operons

• Eukaryotes have more TFs per gene

• Eukaryotes have shorter motifs

• Eukarytes can have long range effects due to DNA folding

• Eukaryotes have non-exact matching mechanism

• Height shows information content at that position ([0,2])

  • If all NUCs are equally represented information is 0

• With small samples (especially if sample number not divisible by 4) a information content of 0 is impossible even though it might be true.

  • pseudocounts try to adress this by adding 1 for each base in all cases

• Different background propabilites

• Regex finds all the patterns, but has no way of distinguishing good from bad hits

• Components:

  • emission propabilities

  • state chagne propabilities

  • states

• The run an hmm is decoded using dynamic programming and the Verterbi algorithm

1. Make an inital guess about the motif

2. Estimate the propability for being the motif at each position in each sequence

3. Compute new propabilites based on the propable hits (factoring hit propablity and nuc frequency)

4. Update the original motiv mode

Initalization is Key, Gibbs sampling tries to use statistical methods to improve EM

1. Translate DNA into all 6 ORFS

2. use gene prediction programm

3. Analyze regulatory sequences

• Consists of proteins and RNA

• Translates mRNA into proteins

• The ribosomal binding site is also transcribed from DNA into mRNA

• The have no introns (uniterupted ORFS)

• They have short intergenic regions

• They have operons ( multiple genes with the same promoter and regulatory sequence)

• They have dense genomes

• Content based (uses bulk properties of seq)

  • e.g ORFS

  • Codon usage

  • Repeat periodicity (less repeats in genes)

• Site based

  • Identification of sites that are typical for genes

  • Donor acceptor sites

  • TF binding sites

  • ATG starts

• comparative

  • Compare the sequence with other genomes and see whether an annoation already exists

• Genes are usually long ORFs

• By change we would expect a stop codon every ~ 21 codons (64/3)

• Short putative genes are propably false positives

• Information that is contained in the sequence

  • Unequall AA usage

  • Condon perferences (different from organism to organism)

  • Unequal codon usage

• HMMs model the grammar of genes

  • e.g every third NUC is the same (AA is mainly encoded by first 2 codon NUCs)

  • assymetry scores are converted to propablilites

• Model are insensitive

• class specific models increase prediction performance

• ORPHEUS

1. DPS local alignments (DPS = DNA protein search)

2. Codon usage analysis (extend local alignments to start/stop condons)

3. RBS weight matrix ( align -20 to -1 regions and identify ribosome binding sites)

4. finde the best gene start

• the ribosomal biding site

• Purine rich region at 5’ end

• AGGAG

• 5-10 bp upstream

• promoter prediction

• splice site prediction

• gene prediction

• One AA is primarily encoded by one codon even though there are multiple others.

• They split wherever they want ( also mid codon)

1. 5’-splice site <exon>GT……<intron>

2. branch point <intron>……A…….<intron>

3. 3’ splice site <intron>…AG<exon>

• The 5’-site is cleaved and loops to the branch point and forms a loop.

• The 3’-site is cleaved

• spliceosome

• Desinged to predict complete gene structures

• based on a general propabilistic model of genomic sequence composition

• based on a GHMM (generalized HMM)

• models 2 strands simultaneously

• Emissions are e.g:

  • N: intergenic region

  • P: promoter

  • E: exon

  • A: poly a tail

• expressed sequence Tag

• expressed RNA the has been retranscribed into CDNA and added into library

• Given a set of putative exons, find the maximun set of non overlapping sequences

• Dynamic programming in O(n)

• Comparisson to EST databases ( are my ESTs already in a database)

• Comparisson of translated seq to known protein sequences

• Comparisson to homologic annotated sequences of related organisims

• CIS: alinging CDNA to its source genome

• TRANS: alinging CDNA to a homologous genome

• The choice of the informant genome

• To similar ( no new information gain )

• To distance ( no simialrities to infer from )

• Ideal seq identiy for informant 55% match

• Sensitivity (How many of a all TPs did I get)

• Specificitiy (Whats the percentage of TP out of all positives)

• Missed Genes ( how many genes didn’t i find)

• Wrong Genes (Where did I predict a gene if there wasn’t any)

• Joined Genes ( did I join 2 genes that are not connected)

• Split Genes ( did I fail to concat 2 exons into one gene)

• Nonfunctional genes, derived from functional genes

• have degenerative features, that prevent expression

• differ from paralogous at crucial points

• Types:

  • conventional

  • processed (at least 8000)

1. Gene duplication

2. One copy is released from evolutionary pressure

3. Aquires lots of mutations

   1. either new gene with similar function

   2. conventional non-functional pseudo gene

• Reinsertion of mRNA into DNA

• Has no:

  1. introns

  2. signaling sequences (TFs, promotors)

• high similarity to func genes can interfere with:

  • PCR

  • in situ hybridization

• Provide a molecular record of genome evolution

• processed / retrotransposed pseudo genes can be used to verify exon struct predictions

• RealTime - PCR using primers that flank AS region

  • run product through gel and check length

• Microarrays

  • probes are exon-exon junctions

  • denpending on which exon-exon junctions a signal is seen the splice product can be infered

• half of all splicing events affect variable regions or entire domains

• other half is non-tirvial

  • affect conserved and strucutred regions

    • => nonsense products

    • => new functions

1. TRUE:

   1. high seq similarity with swiss-prot / trembl seq

2. Putative:

   1. young pseudogenes without frame distruptions

3. Distrupted

• Ka: rate of non-synonymous rate of substitution

• Ks : rate of synonymous substition

• Ka/Ks << 1 “purifying selection”

• Ka/Ks >> 1 “diversify protein product” e.g immune system genes

• Processed pseudo genes Ka/Ks ~1

• positive : binding of an activator enables splicing

  • default no splicing

• negative: binding of an repressor inhibits splicing

  • default splicing

• constitutive:

  • more than one splicing product is always made from a transcribed gene

• regulated:

  • different products are generated at different times, under different condtions

• No some splice sites are only used some of the time

• Retained intron 3%

• Competing 5’-splice sites 18%

  • e.g exon has 2 5’splice sites, only one is used ==> different exon length

• Competing 3’-splice sites 8%

• Exon skipping 38%

• Mutally exclusive exons

• EST libraries

  • better library (coverage) => better prediction

• Splice junction difference (SJD)

• Number of splice junctions that are not on both isoforms / number of all junctions

• evolutionary splicing is conserved

• exons that are not part of constitutive splice forms are mostly not conserved

• SNPs (singel nucleotide polymorphisms)

• Simple Tandom Repeat Polymorphisms

• Insertions / Deletions

• No they occur less often in CDS

• But they can occur anywhere:

  • coding regions

  • introns

  • regulatroy regions

  • intergenic regions

• sequenceing and subsequent alinment to references genome

• Synonymous do not change the AA that is encoded by codon

• Non-synonymous change the AA that is encoded by codon

• experimental charcterization of SNPs is:

  • expensive

  • time consuming

  • difficult

• homology based ( mutations on conserved positions are likely to have a negative effect)

• structure based ( does the mutation meaning fully change the structure)

  
  

• Size change in hydrophobic core ( new AA has larger side chain)

• Introduction of buried charged residues

• protein-protein interaction

• Interferenece with DNA binding

• Mutation of catalytic residues

• SIFT = sorts intorlerant from tolerant

• SIFT does a blast search for a given sequence

• Gets tolerated / deletrious substitions for every pos

• Substitions with norm. props less than a cutoff are decalred deletrious

Fast because it uses homolgy rather than structure

• miRNA : translational regulation

• siRNA : RNA interference

• tRNA : transfer RNA

• Codons don’t recogize the AA they encode

• They recognize the anti codon of a tRNA which brings the AA

• Curcial unpaired regions:

  • 3’end - AA bidning

  • Anticodon loop - Contains anitcodon for matching

• Other features:

  • D loop

  • T loop

• No

• One AA can have mutliple tRNAs

• Some tRNAs can base pair with more than one codon

• Some tRNAs only require accurate matching at the first 2 NUCs of a codon and can wobble on the third

• A-U

• G-C

• G-U (wobble pairing, used to form loop in splicing)

• single stranded

• double stranded

• hairpin loop

• buldge loop

• Pseudo knots

• Evolutionary pressure is on base parinings not sequence

• Methods need to capture primary and secondary information

• Energy minimization

  • pro: accomondation of experimental and alignment data

  • con: not tertiary struct, computationally intensive

• use patterns of covariation

  • pro: simple

  • con: database has to be large enough

• Dynamic programming approach

  • memory intesive

  • might not create most energetically favorable sturct

• familiy of 21-25 small RNAs

• alter the expression of genes in a seq-dependent manner

• pri-miRNA is processed at the Drosha/Pasha complex

• pre-miRNA is exported from nucleus to cytoplasm (Exportin 5)

• pre-miRNA is cleaved by DICER

• one pre-miRNA strand is incorporated into RISC

• Program for identifiying hairpin loops in C.elegans

• How it works:

  • Slide a 110-nt window along genome (discard ovious non miRNA)

  • compute secondary struture for window with RNA fold

  • Identify potential hairpin loop candidates

  • Use scoring criteria on candidates to filter

• multiplicity:

  • one miRNA can traget multiple genes

• cooperativity:

  • one gene can be regulated by many miRNAs

• tandemly repeated DNA

• LINE

• SINE

• Mircrosatelites: Repeats with a repeating sequence between 1-12 bp

  • Can be formed by replication slipage

• Minisatelites: 12-500

• confined to well defined region (e.g telomeric region)

• Can span millions of bp and is species specific

• repeats are believed to play a significant role in genome evolution and disease

• Mobile elements may contain coding regions that are hard to distinguish from other types of genes

• repeats induce local alignments complicating assembly

WTF there is no explanation

Comments:

Anton - hahaha ist Khmer nicht einfach die Suche nach Motiven der Länge k mit n mismatches

Benji - ja aber wan das fürn simpler aaproach

Anton - ich hätte gesagt einfach drüber gehen und schauen welche überrepräsentiert sind für Länge k

• program that uses precompiled libraries to find known repeats

• indispensable in genomes with analyzed repeat families

• new genome => new library

• Class I:

  • first transcribed from DNA

  • reverse transcribed into DNA by TE-encoded RT

  • LTR LINE

• Class II:

  • Move by conservative cut and paste method

• N50:

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

  • e.g 4,641,652

• L50:

  • L50 is the count of smallest number of contigs whose length sum makes up half of genome size.

  • e.g 1

Both are only relevant if there are multiple contigs

• using the wrong codon table

• A software suite for motif:

  • discorvery

  • enrichment

  • scanning

• (counts of G + Counts of C) / seq length

• Yes

• Opposite strand : 3´-AGTCGCAAAGTA-5´

• 5´-ATGAAACGCTGA-3´ ATG … TGA

• family of gene prediction programs

• Used to estimate Codon usage bias quantitaively

• For each AA:

  • for each codon the ratio between itself and most frequent codon is computed

• => most abundant codons have relative adaptiveness of 1

• CAI is geometric mean over weigths associated with each codon

• The gene locations stay the same

• Classes change

• A database for predicted alternate splicing isoforms

• base-pairing propabilities

• breadth: How much of the reference genome was seen during sequenceing

• depth: How often was each Nuc sequenced

1. Genome is broken into chunks of 150000 bp

2. Chunks locations are mapped to genome

3. Chunks are inserted into BACs and grow in bacterial cells

4. Shotgun sequence amplified BACs

• tandemly repeated repeats

• genome wide repeats

• genome sequence variations

• protein-coding genes

• RNA-coding genes

• pseudo genes

• promotors / terminators

• regulatory elements e.g binding motifs

• stochasitc (are DNA and TF close by chance)

• affinity = structural / seq match

• complete match / high affinity is not always desireable

• collection of known DNA motifs from protein sequences

• The propability of a seq x decreases with length

• gets infinitely small

• Log-odds factor in background distributions and compare seq to random seq as background

• Using logs allows for summation

• The sequence is not known

• start is not known

• motifs differ from occurance to occurance

• how do I discren from random motifs

• A seq of nucleotides that occur most often at their given position

• hard to evaluate how good consensus really is

• Gibbs sampler is a stochastic implementation of the EM algorithm

1. Obtain new genomic DNA

2. Translate into all 6 ORFs and compare to protein seq db

   Perform EST db search

3. Use gene prediction programm

4. Analyze regulatory sequence of the gene

• UAA

• UAG

• UGA

• RNA polymerase 1 = rRNA

• RNA polymerase 2 = mRNA

• RNA polymerase 3 = tRNA

• 30-100 FPs for every true one

• include surrounding sequence into prediction

• report everything that is flanked by AG and GT

• The specificity is 0

• Thermodynamic based modelling approach for RNA: RNA duplex interactions

• Input:

  • miRNAs from multiple organisms

  • orthologous 3’ UTR sequences

• translational repression

  • RISC with miRNA binds to mRNA and prevents translation

• mRNA cleavage

  • mRNA is cleaved by RISC

• translational repression is more common

• base pairing porpability sum till 21-nt

• base parining propablitiy sum from 21-nt

• 5’ conversvation

• 3’ conservation

• buldge symetry

• 2300

• tandemly repeated DNA are sequential grouped together repeats

• interspersed repeats are spread accross chromosomes and the entire seq

• Retroelements reproduce by reverse transcription followed by integration into DNA

• DNA transposons are capable of integrating and excising themselves ( cut-and-paste)

• Conservative tranposition ( no copy left behind )

• Replicative transposition ( copy left behind )

• Retrotransposition (RNA intermediary, copy left behind)

• Programm to determine exact repetivie substrings in comple genome

  • Idea: exact matches are core of approximate repeats

• Pro: short running time o(n)

• Con: only exact matches

• given all exact repeats

• define repeat class by merging and extending them

• Seq by sythesis (Ilumina,Roche)

  • Ilumina: shine laser in marked Nuc

  • Roche: pyrophosphate release produces light

• Ligase-mediated sequencing (Applied biosystems)

• Single molecule real Time seuenceing (PAC bio, Oxford nanopore)

  • no library applification

• read quality deteriorates to wards the end of read

• Fixed length cut off

• Adaptive trimming (Quality socre cutoff)

• To distinguish different samples when multiplexing

• Read errors

• Base calling errors

• small insertions / deletions

• YES, DNA motifs are involved in regulating DNA methylation

• BUSCO

• no single reference genome

• A set of reference genomes is a Pan genome

• Unitray Pseudogenes. (deactivated gene, no functional copy)

• Polymorphic ( exhibit variation within population