BIO Assignment Week 2

Assignment for Week 2
Scenario, Labnotes, R-functions, Databases, Data Modeling

Note! This assignment is currently active. All significant changes will be announced on the mailing list.

 
 

Parts labelled as "TBC" are in progress and will be made available as they are being completed.

Concepts and activities (and reading, if applicable) for this assignment will be topics on next week's quiz.

 

The Scenario

I have introduced the concept of "cargo cult science" in class. The "cargo" in Bioinformatics is to understand biology. This includes understanding how things came to be the way they are, and how they work. Both relate to the concept of function of biomolecules, and the systems they contribute to. But "function" is a rather poorly defined concept and exploring ways to make it rigorous and computable will be the major objective of this course. The realm of bioinformatics contains many kingdoms and duchies and shires and hidden glades. To find out how they contribute to the whole, we will proceed on a quest. We will take a relatively well-characterized protein that is part of a relatively well-characterized process, and ask what its function is. We will examine the protein's sequence, its structure, its domain composition, its relationship to and interactions with other proteins, and through that paint a picture of a "system" that it contributes to.

Our quest will revolve around a transcription factor that plays an important role in the regulation of the cell cycle: Mbp1 is a key component of the MBF complex (Mbp1/Swi6) in yeast. This complex regulates gene expression at the crucial G1/S-phase transition of the mitotic cell cycle and has been shown to bind to the regulatory regions of more than a hundred target genes. It is therefore a DNA binding protein that acts as a control switch for a key cellular process.

We will start our quest with information about the Mbp1 protein of Baker's yeast, Saccharomyces cerevisiae, one of the most important model organisms. Baker's yeast is a eukaryote that has been studied genetically and biochemically in great detail for many decades, and it is easily manipulated with high-throughput experimental methods. But each of you will use this information to study not Baker's yeast, but a related organism. You will explore the function of the Mbp1 protein in some other species from the kingdom of fungi, whose genome has been completely sequenced; thus our quest is also an exercise in model-organism reasoning: the transfer of knowledge from one, well-studied organism to others.

It's reasonable to hypothesize that such central control machinery is conserved in most if not all fungi. But we don't know. Many of the species that we will be working with have not been characterized in great detail, and some of them are new to our class this year. And while we know a fair bit about Mbp1, we probably don't know very much at all about the related genes in other organisms: whether they exist, whether they have similar functional features and whether they might contribute to the G1/S checkpoint system in a similar way. Thus we might discover things that are new and interesting. This is a quest of discovery.

Here are the steps of the assignment for this week:

1. We'll need to explore what data is available for the Mbp1 protein.
2. We'll need to pick a species to adopt for exploration.
3. We'll need to define what data we want to store and design a datamodel.

However, before we head off into the Internet: have you thought about how to document such a "quest"? How will you keep notes? Obviously, computational research proceeds with the same best-practice principles as any wet-lab experiment. We have to keep notes, ensure our work is reproducible, and that our conclusions are supported by data. I think it's pretty obvious that paper notes are not very useful for bioinformatics work. Ideally, you should be able to save results, and link to files and Webpages.

 

Keeping Labnotes

Consider it a part of your assignment to document your activities in electronic form. Here are some applications you might think of - but (!) disclaimer, I myself don't use any of these (yet) (except the Wiki of course).

• Evernote - a web hosted, automatically syncing e-notebook.
• Nevernote - the Open Source alternative to Evernote.
• Google Keep - if you have a Gmail account, you can simply log in here. Grid-based. Seems a bit awkward for longer notes. But of course you can also use Google Docs.
• Microsoft OneNote - this sounds interesting and even though I have had my share of problems with Microsoft products, I'll probably give this a try. Syncing across platforms, being able to format contents and organize it sounds great.
• The Student Wiki - of course. You can keep your course notes with your User pages.

Are you aware of any other solutions? Let us know!

Keeping such a journal will be helpful, because the assignments are integrated over the entire term, and later assignments will make use of earlier results. But it is also excellent practice for "real" research. Expand the section below for details - written from a Wiki perspective but generally applicable.

 

Remember you are writing a lab notebook—not a formal lab report: a point-form record of your actual activities. Write such documentation as notes to your (future) self.

Create a lab-notes page as a subpage of your User space on the Student Wiki.

For each task:

• Write a header and give it a unique number.

This is useful so you can refer to the header number in later text. Obviously, you should "hard-code" the number and not use the Wiki's automatic section numbering scheme, since the numbers should be stable over time, not change when you add or delete a section. It may be useful to add new contents at the top, so you don't have to scroll to the bottom of the page evry time you add new material. This does not have to be in strict chronological order, like we would have it in a paper notebook. It may be advantageous to give different subprojects their own page, or at least order them on one page. Just remember that things that are on the same page are easy to find.

• State the objective.

In one brief sentence, restate what your task is supposed to achieve.

• Document the procedure.

Note what you have done, as concisely as possible but with sufficient detail. I am often asked: "What is sufficient detail"? The answer is easy: detailed enough so that someone can reproduce what you have done. In practice that guy will often be you, yourself, in the future. I hope that you won't be constantly cursing your past-self because of omissions!

• Document your results.

You can distinguish different types of results -

• • Static data does not change over time and it may be sufficient to note a reference to the result. For example, there is no need to copy a GenBank record into your documentation, it is sufficient to note the accession number or the GI number, or better, to link to it.
  • Variable data can change over time. For example the results of a BLAST search depend on the sequences in the database. A list of similar structures may change as new structures get solved. In principle you want to record such data, to be able to reproduce at a later time what your conclusions were based on. But be selective in what you record. For example you should not paste the entire set of results of a BLAST search into your document, but only those matches that were important for your conclusions. Indiscriminate pasting of irrelevant information will make your notes unusable.
  • Analysis results

The results of sequence analyses, alignments etc. in general get recorded in your documentation. Again: be selective. Record what is important.

• Note your conclusions.

An analysis is not complete unless you conclude something from the results. (Remember what we said about "Cargo Cult Science". If there is no conclusion, your activities are quite pointless.) Are two sequences likely homologues, or not? Does your protein contain a signal-sequence or does it not? Is a binding site conserved, or not? The analysis provides the data. In your conclusion you provide the interpretation of what the data means in the context of your objective. Were you expecting a signal-sequence but there isn't one? What could that mean? Sometimes your assignment task in this course will ask you to elaborate on an analysis and conclusion. But this does not mean that when I don't explicitly mention it, you can skip the interpretation.

• Add cross-references.

Cross-reference to other information are super valuable as your documentation grows. It's easy to see how to format a link to a section of your Wiki-page: just look at the link under the Table of Contents at the top. But you can also place "anchors" for linking anywhere on an HTML page: just use the following syntax. <span id="{some-label}"><\span> for the anchor, and append #{some-label} to the page URL. Try this here: (http://steipe.biochemistry.utoronto.ca/abc/Assignment_2#tf) .

• Use discretion when uploading images

I have enabled image uploading with some reservations, we'll see how it goes. You must not:

• upload images that are irrelevant for this course;
• upload copyrighted images;
• upload any images that are larger than 500 kb. I may silently remove large images when I encounter them.

Moreover, understand that any of your uploaded images may be deleted at any time. If they are valuable to you, keep backups on your own machine.

• Prepare your images well

Don't upload uncompressed screen dumps. Save images in a compressed file format on your own computer. Then use the Special:Upload link in the left-hand menu to upload images. The Wiki will only accept .jpeg or .png images.

• Use the correct image types.

In principle, images can be stored uncompressed as .tiff or .bmp, or compressed as .gif or .jpg or .png. .gif is useful for images with large, monochrome areas and sharp, high-contrast edges because the LZW compression algorithm it uses works especially well on such data; .jpg (or .jpeg) is preferred for images with shades and halftones such as the structure views you should prepare for several assignments, JPEG has excellent application support and is the most versatile general purpose image file format currently in use; .tiff (or .tif) is preferred to archive master copies of images in a lossless fashion, use LZW compression for TIFF files if your system/application supports it; The .png format is an open source alternative for lossless, compressed images. Application support is growing but still variable. .bmp is not preferred for really anything, it is bloated in its (default) uncompressed form and primarily used only because it is simple to code and ubiquitous on Windows computers.

Image dimensions and resolution

Stereo images should have equivalent points approximately 6cm apart. It depends on your monitor how many pixels this corresponds to. The dimensions of an image are stated in pixels (width x height). My notebook screen has a native display resolution of 1440 x 900 pixels/23.5 x 21 cm. Therefore a 6cm separation on my notebook corresponds to approximately 260 pixels. However on my desktop monitor, 260 pixels is 6.7 cm across. And on a high-resolution iPad display, at 227 ppi (pixels per inch), 260 pixels are just 2.9 cm across. For the assignments: adjust your stereo images so they are approximately at the right separation and are approximately 500 to 600 pixels wide. Also, scale your molecules so they fill the available window and - if you have depth cueing enabled - move them close to the front clipping plane so the molecule is are not just a dim blob, lost in murky shadows.

Considerations for print (manuscripts etc.) are slightly different: for print output you can specify the output resolution in dpi (dots per inch). A typical print resolution is about 300 dpi: 6 cm separation at 300dpi is about 700 pixels. Print images should therefore be about three times as large in width and height as screen images.

Preparation of stereo views

When assignments ask you to create molecular images, always create stereo views.

Keep your images uncluttered and expressive

Scale the molecular model to fill the available space of your image well. Orient views so they illustrate a point you are trying to make. Emphasize residues that you are writing about with a contrasting colouring scheme. Add labels, where residue identities are not otherwise obvious. Turn off side-chains for residues that are not important. The more you practice these small details, the more efficient you will become in the use of your tools.

If you have technical difficulties, post your questions to the list and/or contact me.

 

Data Sources

SGD - a Yeast Model Organism Database

Yeast happens to have a very well maintained model organism database - a Web resource dedicated to Saccharomyces cerevisiae. Where such resources are available, they are very useful for the community. For the general case however, we need to work with one of the large, general data providers - the NCBI and the EBI. But in order to get a sense of the type of data that is available, let's visit the SGD database first.

Task:
Access the information page on Mbp1 at the Saccharomyces Genome Database.

1. Browse through the Summary page and note the available information: you should see:
   • information about the gene and the protein;
   • Information about it's roles in the cell curated at the Gene Ontology database;
   • Information about knock-out phenotypes; (Amazing. Would you have imagined that this is a non-essential gene?)
   • Information about protein-protein interactions;
   • Regulation and expression;
   • A curators' summary of our understanding of the protein. Mandatory reading.
   • And key references.
2. Access the Protein tab and note the much more detailed information.
   • Domains and their classification;
   • Sequence;
   • Shared domains;
   • and much more...

You will notice that some of this information relates to the molecule itself, and some of it relates to its relationship with other molecules. Some of it is stored at SGD, and some of it is cross-referenced from other databases. And we have textual data, numeric data, and images.

How would you store such data to use it in your project? We will work on this question at the end of the assignment.

 

---

 

If we were working on yeast, most data we need is right here: curated, kept current and consistent, referenced to the literature and ready to use. But you'll be working on a different species and we'll explore the much, much larger databases at the NCBI for this. The upside is that most of the information like this is available for your species. The downside is that we'll have to integrate information from many different sources "by hand".

 

NCBI databases

The NCBI (National Center for Biotechnology Information) is the largest international provider of data for genomics and molecular biology. With its annual budget of several hundred million dollars, it organizes a challenging program of data management at the largest scale, it makes its data freely and openly available over the Internet, worldwide, and it runs significant in-house research projects.

Let us explore some of the offerings of the NCBI that can contribute to our objective of studying a particular gene in an organism of interest.

 

Entrez

Task:
Remember to document your activities.

1. Access the NCBI website at http://www.ncbi.nlm.nih.gov/
2. In the search bar, enter mbp1 and click Search.
3. On the resulting page, look for the Protein section and click on the link. What do you find?

The result page of your search in "All Databases" is the "Global Query Result Page" of the Entrez system. If you follow the "Protein" link, you get taken to the more than 450 sequences in the NCBI Protein database that contain the keyword "mbp1". But when you look more closely at the results, you see that the result is quite non-specific: searching only by keyword retrieves an Arabidopsis protein, bacterial proteins, a Saccharomyces protein (perhaps one that we are actually interested in), Maltose Binding Proteins - and much more. There must be a more specific way to search, and indeed there is. Time to read up on the Entrez system.

Task:

1. Navigate to the Entrez Help Page and read about the Entrez system, especially about:
   1. Boolean operators,
   2. wildcards,
   3. limits, and
   4. filters.
2. You should minimally understand:
   1. How to search by keyword;
   2. How to search by gene or protein name;
   3. How to restrict a search to a particular organism.

Don't skip this part, you don't need to know the options by heart, but you should know they exist and how to find them. It would be great to have a synopsis of the important fields for reference, wouldn't it?Why don't you go and create one: I have put a template page on the Student Wiki (A synopsis of Entrez codes). Editors welcome!

Keyword and organism searches are pretty universal, but apart from that, each NCBI database has its own set of specific fields. You can access the keywords via the Advanced Search interface of any of the database pages.

 

Protein Sequence

 

Task:
With this knowledge we can restrict the search to proteins called "Mbp1" that occur in Baker's Yeast. Return to the Global Search page and in the search field, type:

Mbp1[protein name] AND
 "Saccharomyces cerevisiae"[organism]

This should find one and only one protein. Follow the link into the protein database: since this is only one record, the link takes you directly to the result: CAA98618.1—a data record in Genbank Flat File (GFF) format. The database identifier CAA98618.1 tells you that this is a record in the GenPept database. There are actually several, identical versions of this sequence in the NCBI's holdings. A link to "Identical Proteins" near the top of the record shows you what these are:

Some of the sequences represent duplicate entries of the same gene (Mbp1) in the same strain (S288c) of the same species (S. cerevisiae). In particular:

• there are three GenBank; records (CAA52271.1, CAA98618.1 and DAA11800.1); these are archival entries, submitted by independent yeast genome research projects;

• there is an entry in the RefSeq database: NP_010227.1. This is the preferred entry for us to work with. RefSeq is a curated, non-redundant database which solves a number of problems of archival databases. You can recognize RefSeq identifiers – they always look like NP_12345.1, NM_12345.1, XP_12345.1, NC_12345.1 etc. This reflects whether the sequence is protein, mRNA or genomic, and inferred or obtained through experimental evidence. The RefSeq ID NP_010227.1 actually appears twice, once linked to its genomic sequence, and once to its mRNA.

• there is a SwissProt sequence P39678.1. This link is kind of a big deal. It's a cross-reference into UniProt, the huge protein sequence database maintained by the EBI (European Bioinformatics Institute), which is the NCBI's counterpart in Europe. SwissProt entries have the highest annotation standard overall and are expertly curated. Many Webservices that we will encounter, work with UniProt ID's (e.g. P39678.1), rather than RefSeq. But it used to be until recently that the two databases did not link to each other, mostly for reasons of funding politics. It's great to see that this divide has now been overcome.

• Finally, there are four entries of the same sequence in different yeast strains. These don't have to be identical, they just happen to be. Sometimes we find identical sequences in quite divergent species. Therefore I would not actually consider EIW11153.1, AJU86440.1, AJU58508.1, and AJU61971.1 to be identical proteins, although they have the same sequence.

Note all the .1 suffixes of the sequence identifiers. These are version numbers. Two observations:

1. It's great that version numbers are now used throughout the NCBI database. This is good database engineering practice because it's really important for reproducible research that updates to database records are possible, but recognizable. When working with data you always must provide for the possibility of updates, and manage the changes transparently and explicitly. Proper versioning should be a part of all datamodels.
2. When searching, or for general use, you should omit the version number, i.e. use NP_010227 or P39678 not NP_010227.1 resp. P39678.1. This way the database system will resolve the identifier to the most current, highest version number (unless you want the older one, of course).

Task:

1. Note down the RefSeq ID and the UniProt (SwissProt) ID in your journal.
2. Follow the link to the RefSeq entry NP_010227.1.
3. Explore the page and follow these links (note the contents in your journal):
   1. Under "Analyze this Sequence": Identify Conserved Domains
   2. Under "Protein 3D Structure": See all 3 structures...
   3. Under "Pathways for the MBP1 gene": Cell cycle - yeast
   4. Under "Related information" Proteins with Similar Sequence

As we see, this is a good start page to explore all kinds of databases at the NCBI via cross-references.

 

PubMed

Arguably one of the most important databases in the life sciences is PubMed and this is a good time to look at PubMed in a bit more detail.

Task:

1. Return back to the MBP1 RefSeq record.
2. Find the PubMed links under Related information in the right-hand margin and explore them.
   1. The first one (PubMed) will take you to records that cite the sequence record;
   2. The second one (PubMed (RefSeq)) will take you to articles that relate to the Mbp1 gene or protein;
   3. The third one (PubMed (Weighted)) applies a weighting algorithm to find broadly relevant information - an example of literature data mining. PubMed(weighted) appears to give a pretty good overview of systems-biology type, cross-sectional and functional information.

But neither of the searches finds all Mbp1 related literature.

1. On any of the PubMed pages open the Advanced query page and study the keywords that apply to PubMed searches. These are actually quite important and useful to remember. Make yourself familiar with the section on Search field descriptions and tags in the PubMed help document, (in particular [DP], [AU], [TI], and [TA]), how you use the History to combine searches, and the use of AND, OR, NOT and brackets. Understand how you can restrict a search to reviews only, and what the link to Related citations... is useful for.
2. Now find publications with Mbp1 in the title. In the result list, follow the links for the two Biochemistry papers, by Taylor et al. (2000) and by Deleeuw et al. (2008). Download the PDFs, we will need them later.

Now, we were actually trying to find related proteins in a different species. Our next task is therefore to decide what that species should be.

 

Choosing YFO (Your Favourite Organism)

TBC

Data modelling

TBC

 

That is all.

 

Links and resources

 

Footnotes and references

 

Ask, if things don't work for you!

If anything about the assignment is not clear to you, please ask on the mailing list. You can be certain that others will have had similar problems. Success comes from joining the conversation.

• Do consider how to ask your questions so that a meaningful answer is possible:
  • How to create a Minimal, Complete, and Verifiable example on stackoverflow and ...
  • How to make a great R reproducible example are required reading.