Bio 123, UZH

E: Sequenced reads that overlap are reassembled back into longer sequences.

D: Ein Contig ist ein Satz überlappenderDNA- oder Protein-Stücke (reads), die von derselben genetischen Quelle stammen. Ein solches Contig kann dazu genutzt werden, die Original-DNA-Sequenz dieser genetischen Quelle abzuleiten.

A series of linked contig which are in the right order but not necessary connected in one continuous stretch of sequence

Either how many or how long reads we create

The average number of reads representing a given nucleotide in the reconstructed sequence is called depth of coverage

See picture

- High number of reads and low input

- Short read but very high throughput

- clonal amplification

- State-of-the-arts optics used for detection

- At the end of the bridge amplification, you get a cluster. This is what you actually going to sequence!

- SBS (sequencing by synthesis)

- Mid-to-High number of reads and low input

- Short reads and low throughput

- but very quick (usefull during a surgery when quick results are needed)

- beads (Kugel) are used for monoclonal amplifications

- ph-meter is used

- no modified nucleotides

- Low number of reads and high input

- SMRT (single molecule real time sequencing)

- Eavesdropping (lauschen) on the DNA polymerase as it replicates DNA 

- Only to the last phosphore ist fluorescent dye attache, the strand is natural

-> https://www.youtube.com/watch?v=v8p4ph2MAvI

- Low numper of reads and high input

- sequencing of long fragments

- flow of ions that are transported through perforated membranes

- https://www.youtube.com/watch?v=3UHw22hBpAk

1.       Forward Genetics

This is the more classical approach. Here you do random mutagenesis and then screen for cells with a change in the phenotype. So you select for a phenotype, and then identify the mutated gene.

2.       Reverse Genetics

This is really the opposite, I already know which gene I want to test, and what mutant, but in the past we couldn’t really do that because we didn’t know all genes.

So here you start with a gene of unknown function (but you know exactly which gene it is), then you do (only with this gene!) an overexpression or another sort of mutation. Then you examine the resulting change in the phenotype to infer the function.

-> "targeted" mutagenesis

- Grow a very large number of cells in a dish,

- Then you throw in an agent who randomly mutate some cells, but limited so you get one or even less mutation per cell! (=limited random mutagenesis)

- Next step is to select the cells with the desired phenotyp. There are two ways you can do that:

Either by testing the resistance (throw something on the cell that kill all the cells BUT the one with mutations, for example toxic compound or a virus.) or second possibility by a fluorescent reportert (you can mark either type, the normal ones or the ones with a mutation)

The last step is to identify the mutation by sequencing.

-First generate a large number of reagents each perturbing a specific gene. (3 important methods: cDNA, RNAi, CRISPR-CAS9)

- Then make an arrayed (geordnet) library where each reagent has a pre-defined position (which is called a "well"). Third step is to add cells to each well and systematically score for phenotypes. Last, you assign the function to each gene.

cDNA = copy DNA from mRNA

You use reverse transcriptase to get a DNA from a mRNA. Because the mRNA is already processed (e.g. Spliced) there are also no introns etc.

With miRNA and RNA interference you can find other RNA or even degrade the whole RNA in a cell. -> Knock down

An adaptive immune systeme of bacteria adapted to work in any cell

It allows them to make transcripts with a little seqeuence of this foreign DNA (of a virus), so when this virus come again at a second infection, the bacteria is now able to recognize and cleave it -> knock out

A single guide RNA you can design some that can targeting any gene you want

You introduce a few mutations, cell try to repair that, but then often you get a frame shift, so you’re gene is not functional (functional knock out)

Allg. Erklärung auf Deutsch: http://www.transgen.de/lexikon/1845.crispr-cas.html ->

Finden: Der CRISPR-Abschnitt erkennt mit Hilfe der darin integrierten RNA (Guide RNA) das jeweilige Ziel, eine bestimmte Sequenz in dem umzuschreibenden Gen.

Schneiden: Das an den CRISPR-Abschnitt gekoppelte Cas9-Protein schneidet den DNA-Doppelstrang genau an der gewünschten Zielsequenz. Beide Elememte - CRISPR und Cas9 - werden synthetisch hergestellt und anschließend in eine Zelle eingeführt.

Reparieren: Die zelleigenen Reparatursysteme fügen nun den durchtrennten DNA-Strang wieder zusammen. Dabei können einzelne DNA-Bausteine abgeschaltet oder wieder aktiviert werden. Möglich ist auch, kurze DNA-Sequenzen neu einzubauen

Zernike function can be used to describe the place of the proteins

1. Segment Cell and nucleus -> with segmentation you can quantify all kinds of properties from each cell 

2. Quantitative read-outs of phenotypes ->How does the phenotyp change after a pertubation?

3. Genes with similar loss-of-function can be connected in networks -> Functional modulses

- Catalytic (enzymes, e.g. kinases)

- Transport (hemoglobin)

- Structural (keratins)

- Hormonal (insulin)

- Immunity (antibodies)

The total protein complement of a genome or all proteins of the cells present at any given time.

Proteomics is the study of protein properties on a large scale to obtain global, integrated view of disease processes , cellular processes and networks at the protein level. They are dynamic!

Some examples for properties of proteins are expression-level, modification, interaction, structure, location, etc.

Proteins are organized in networks. It’s a set of nodes and edges. Edges describe a relationship between the nodes. In any modell, you have always to define what these edges stand for (does it regulate? Or binds?).

- Path length (Thes shortest path between two nodes)

- Network Diameter (the longest shortest path)

- Degree k_i (The number of edges involving node i)

- Degree distribution P(k) (The probability (frequency) of nodes of degree k)

- Clustering Coefficient

see picture

see picture

A scale free network is when you have  some nodes that are very highly connected, and many which are only connected to some//the majority often have one edge, only a small fraction have a hundred edges -> power law degree distribution

(the highly connected ones are called hubs) like airports

The big advantage of them is that they are resistant to component failure. And the topology is important for robustness.

And the big disadvantage is of course its attack vulnerability. If the hubs has a problem, the whole network could collapse.  

The hubs doesn't tend to interact directly with each other, and they also tend to be "older" proteins. Hubs also seem to be more conserved than average between species. (because there is a lot of selective pressure!)

The density of the network

This has to do with the complex structure and the big variety of proteins.  (humans: 20`000) and also the huge abundance range. For example the ones that are present in one cell in 10^6copies, we find them easily. But what if there are only  <50 copies present?

• Colorimetric
• UV spectroscope 
• Fluorescent-based
• Antibodies
• Mass spectometry

Mass Spectrometer measures mass over charge ratio (m/z) of ions. 

m=Masse

z= Ladungsverhältnis

The MS and the MS/MS mode.

The MS/MS works with isolation, where you only work with one peptid at the end.

The MS first ionize the sample, then analyze and then detect it. In the end you get a data analysis.

1.       Identification of system components

A.      Gel-based approaches (but very obsolete)

B.      Shotgun proteomics

You shoot but you don’t know what you shoot exactly.

C.      Targeted proteomics

You use when you’re interested in a specific protein, which you already know before the experiment. (Quantification of a specific list of proteins across many perturbations)

The advantages are its multicomplexed, it distinguishes isoforms and it’s precise and reproducible. But there is tedious assay development and difficult implemenation for a non-specialized lab

2.       Component abundance changes

If you want to know what is changing in a network during time, you have to follow a network over a time.

Here you use quantitative proteomics. It is done by comparing the signal intensity of peptides with the same sequence, but different isotop labels.

3.       Absolute quantification of a network

F.e reaction rates, stochiometry, stochastity, ...

Here you choose very informative peptides

(from which you know very accurately the concentrations!!!

So you can compare it with others)

4.       Modifications

5.       Prot-Prot interactions

6.       Structural networks

A) Data Quality

• Completeness
• Reproducibility
• Measurement Accuracy
• Biological Noise

B) Data Representation

• Visualization
• Patterns/Trends

C) Data Interpretation

• False Discovery RAte
• Signal vs. Noise
• Multiple Testing
• Modelling

• Network Topology
• Network Motifs
• Biological Insights from Topology
• Network Partitioning/Clustering
• Networks as Tools for Data Representation