BCH394P BCH364C 2023
BCH394P/BCH364C Systems Biology & Bioinformatics
Course unique #: 55425/55330
Lectures: Tues/Thurs 11 – 12:30 PM WEL 2.110
Instructor: Edward Marcotte, marcotte @ utexas.edu
- Office hours: Mon 4 – 5 PM on the class Zoom channel (available on Canvas)
TA: Matt McGuffie, mmcguffie @ utexas.edu
- TA Office hours: Wed 1 - 2 PM / Thu 12:30 - 1:30 in MBB 1.448BA or by appointment on Zoom
Class Canvas site: https://utexas.instructure.com/courses/1352289
Lectures & Handouts
Mar 14,16, 2023 - SPRING BREAK
- Don't forget to turn in the proposal for your course project by March 21st and finish HW3 by March 22nd.
Mar 9, 2023 - Clustering II
- We'll be continuing the slides from last time
- I'm also posting the next (last) problem set:
Problem Set 3, due before 10PM Mar. 22, 2023. You will need the following software and datasets:
- The clustering software is available here. There is an alternative package here that you can download and install on your local computer if you prefer.
- Amino acid sequences of 1832 human proteins
- Human protein phylogenetic profiles. These data come from this paper.
- Human protein co-fractionation/mass spectrometry profiles. These data come from this paper.
- Review of phylogenetic profiles
- Fuzzy k-means
- SOM gene expression
- Links to various applications of SOMs: 1, 2, 3. You can run SOM clustering with the Open Source Clustering package with the '-s' option, or GUI option (here's the manual). (FYI, it also supports PCA). If you are not happy with Cluster's SOM function, the statistical package R also provides a package for calculating SOMs (http://cran.r-project.org/web/packages/som/index.html).
- t-SNE and UMAP
- Links to various applications of t-SNE: 1, 2, 3, 4. You can run t-SNE and UMAP on the following web site.
Mar 7, 2023 - Functional Genomics & Data Mining - Clustering I
- Due March 21 by email to the TA+Instructor - One to two (full) paragraphs describing your plans for a final project, along with the names of your collaborators. Please limit to no more than 3 per group, please. It's also fine to do this independently, if you prefer. (Do you have a particular skill/interest/exciting dataset you need help analyzing? There is a class_projects channel on the slack where you can ask around for partners.) This assignment (planning out your project) will account for 5 points out of your 25 total points for your course project. Here are a few examples of final projects from previous years: 1 2 3 4 5 6 7 8 9 10 11 12 13 14
- Science news of the day: The genome of Antarctic krill (the crustacean E. superba) has been sequenced and is crazy. It's 48 Gb in size, so 15x the human genome (!), one of the largest genomes ever assembled. And >92% of that is repetitive DNA. Solved with a combination of short and long read DNA sequencing.
- Today's slides
- Primer on clustering
- K-means example (.ppt)
- Here's a nice explanation of some of the various distance measures used for clustering
- B cell lymphomas
Mar 2, 2023 - Motifs
- We'll talk about motif finding today.
- Today's slides
- Wordle as an excuse to learn about information theory & entropy and sequence logos and motifs!
- NBT Primer - What are motifs?
- NBT Primer - How does motif discovery work?
- The biochemical basis of a particular motif
- Gibbs Sampling
Feb 28, 2023 - NGS analysis best practices
- Homework #3 (worth 10% of your final course grade) has been assigned on Rosalind and is due by 10:00PM March 9. In past years, we've run into problems with Rosalind timing out before Meme completes although it usually runs eventually, so be warned you may have to try it a couple of times. Meme also runs faster using the "zero to one" or "one" occurrence per sequence option, rather than the "any number of repeats" option.
- Guest speaker: Anna Battenhouse from the Center for Biomedical Research Support, where she maintains the Biomedical Research Computing Facility.
- Today's slides
Feb 23, 2023 - Genome Assembly/Mapping II
- We're finishing up the slides from last time. Note that we give short shrift to read mapping/alignment algorithms, of which there are now a very long list. Here's an interesting discussion by Lior Pachter of the major developments in that field.
- Here is an excellent explanation (now archived) of how the BWT relates to a suffix tree and enables fast read mapping to a genome
- If you want a more detailed explanation, the BWA paper more formally describes how the Burrows–Wheeler transform can be used to construct an index.
- Two notable advances in genome assembly: String Graphs and more recently, multiplexed De Bruijn graphs. Both have been used to assemble a fully complete human genome sequence (check out the beautiful string graph visualizations of the final assemblies, which capture gapless telomere-to-telomere assemblies for all 22
human autosomes and Chromosome X)
Feb 21, 2023 - Genome Assembly
- Science news of the day: New evidence for very short human ORFs coding for real microproteins & peptides
- & A compilation of advances in the last 2 years on deep learning protein structure prediction
- Relevant to the last lecture, some definitions of sensitivity/specificity & precision/recall. Note that the gene finding community settled early on to a different definition of specificity that corresponds to the precision or PPV in other fields. Other fields define specificity as the true negative rate.
- Today's slides
- DeBruijn Primer and Supplement
PROBLEM SET #2 ANNOUNCEMENT
- If you would like a few examples of proteins annotated with their transmembrane and soluble regions (according to UniProt) to help troubleshoot your homework, here are some example yeast protein sequences.
Feb 16, 2023 - Gene finding II
- Short classes at UT start this week in genome sequencing, proteomics, and bioinformatics
- Several of you have asked about programming the Viterbi algorithm for the homework, so I wanted to make sure everyone realized that you are not required to program it. The sequence is short enough that you can solve it in a spreadsheet if that's easier for you.
- We're finishing up the slides from last time.
- Reposting this so it doesn't fall through the cracks: The current state of gene annotation
- Why do we call it the Viterbi algorithm?
Feb 14, 2023 - Gene finding
- Happy Valentine's Day!
- Today's slides on gene finding
- A nice commentary on gene finding: Next-generation genome annotation: we still struggle to get it right
- For a few more examples of HMMs in action, here's a paper on sequencing the human genome by nanopore, which used HMMs in 3-4 different ways for polishing, contig inspection, repeat analysis and 5-methylcytosine detection.
- The UCSC genome browser
- A few useful links about programming: Recommendations for "good enough" programming habits and a great Python beginners Youtube tutorial
Reading (a couple of old classics + a review + better splice site detection):
Feb 9, 2023 - HMMs II
- Science news of the day: a fun preprint illustrating the scale of efforts to identify protein families. This one clustered "19 billion sequences in 18 days on 27 high performance computing nodes, using 250,000 CPU hours in total". In all, they found 544 million sequence families (clusters) capturing ~94% of all known proteins, giving a sense of the overall size of the universe of proteins.
Problem Set 2, due before 10 PM, Feb. 20, 2023:
- Problem Set 2.
- You'll need these 3 files: State sequences, Soluble sequences, Transmembrane sequences
- Link to a great interactive visualization of Markov chains, by Victor Powell & Lewis Lehe. It's worth checking out to build some intuition. They correctly point out that Google's PageRank algorithm is based on Markov chains. There, the ranking of pages in a web search relates to how random walks across linked web pages spend more time on some pages than on others.
- A non-biological example of using log odds ratios & Bayesian stats to learn the authors of the Federalist Papers. In a related example, researchers just decoded >50 coded letters from a French archive and discovered they were lost correspondence from Mary, Queen of Scots, before she was executed in 1587 for treason against Elizabeth I. The researchers used an approach closely related to computing log odds ratios of 5-mer frequencies between putative decoded texts and known free text to figure out the correct ciphers. If you're curious, you can read about it in Appendix A of their paper
Feb 7, 2023 - Hidden Markov Models
- Don't forget: Rosalind Homework #2 (worth 10% of your final course grade) is due by 10 PM February 8. Note: choose one of the two protein translation problems and see the update below on the IUPAC code example.
- More stats for comp biologists worth checking out: Modern Statistic for Modern Biology, by Susan Holmes and Wolfgang Huber. It's currently available online and available on dead tree. (FYI, all code is in R.)
- Today's slides
- HMM primer and Bayesian statistics primer #1, Bayesian statistics primer #2, Wiki Bayes
- Care to practice your regular expressions? (In python? & a Python regexp cheat sheet)
- It looks like a number of people are struggling with the Rosalind problem titled Protein Translation. As an alternative option, I've assigned a problem titled Translating RNA into Protein. Choose one; you'll get credit regardless of which of them you do. Also, it looks like the problem titled "Complementing a Strand of DNA" uses a now out-of-date call for IUPAC codes in the Programming Shortcut. Just delete the "from Bio.Alphabet import IUPAC" line & delete the ", IUPAC.unambiguous_dna" portion of the Seq() functions and it should work fine.
Feb 2, 2023 - We'll have a guest lecture from your TA Matt McGuffie on advancing your Python data analysis skills
- WEATHER WARNING #2: Change of plans! UT has now officially canceled in-person classes, but more to the point, >100,000 people have lost power in Austin today. We're going to cancel the live zoom class tomorrow, and Matt will instead record the lecture and upload it to Canvas for viewing.
- Matt is an expert in the bioinformatic analyses of plasmid sequences and developed the popular pLannotate tool to annotate and visualize plasmid features, based on a large database of genetic parts and protein sequences. Funny enough, he first described an early version of pLannotate as his project for this class back in 2019. He'll be introducing several useful Python libraries, including the Pandas package for handling large tables and a data visualization library for plotting data.
Jan 31, 2023 - Biological databases
- WEATHER WARNING: UT just announced a campus closure for the morning, so for those of you that are able to attend online, I'll plan to hold it at the normal time on the class zoom channel (link available on Canvas). However, for those that can't make it, don't stress! We'll record the lecture and post the video to Canvas so that you can watch it later. Note: the next Rosalind homework is assigned below.
- Science news of the day: Cell, Nature, Science, eLife, and the Lancet ban listing ChatGPT as a co-author
- Today's slides
Homework #2 (worth 10% of your final course grade) has been assigned on Rosalind and is due by 10 PM February 8:
- Besides giving a bit more programming experience, these questions will also introduce you to the BioPython Python library (see the "programming shortcuts" at the bottom of several questions). If you need to install BioPython on your computer, open an Anaconda prompt window (on a PC) or launch a console window from the Anaconda Navigator & type "pip install biopython". (You can use this approach to install most Python libraries.) There's a very useful tutorial here (also downloadable as a pdf file)
- Just a note that we'll be seeing ever more statistics as go on. Here's a good primer from Prof. Lauren Ancel Myers (who leads the UT Austin COVID-19 Modeling Consortium) to refresh/explain basic concepts.
- Finally, here's great opportunity to hone your Python skills a bit more: The UT CBRS cores will offer short courses in Python, Unix, and Python for Data Sciences starting in March.
Jan 26, 2023 - BLAST
- Our slides today are modified from a paper on Teaching BLAST by Cheryl Kerfeld & Kathleen Scott.
- The original BLAST paper
- The protein homology graph paper. Just for fun, here's a stylized version of this plot that we exhibited in the engaging Design and the Elastic Mind show at New York's Museum of Modern Art, now in their permanent collection.
Jan 24, 2023 - Sequence Alignment II
- We'll be finishing up slides from last time.
- Problem Set 1 clarification: for problems asking for "nucleotide frequencies" = turn in the absolute count of each nucleotide (or dinucleotide) as well as the fractions or percentages of the total
- Science news of the day: We're about 3 years from publication of the SARS-CoV-2 genome papers 1 2. The release of the genome sequences immediately launched the COVID vaccine design process. Here's a great write-up in the NYT of the story of the vaccine development process, including the McLellan lab's key S2P double proline mutations introduced to stabilize the spike protein. It was just selected as one of the winners of the USPTO Patents for Humanity award in the COVID-19 category
- For those of you who could use more tips on programming, the weekly peer-led open coding hour is starting up again! Every Monday, 3:30-4:30, in the MBB 2.232 lounge. It's a very informal setting where you can work and ask questions of more experienced programmers.
- Fact and Fiction in Sequence Alignments
- Dynamic programming primer
- An example of dynamic programming using Excel, created by Michael Hoffman (a former U Texas undergraduate, now U Toronto professor, who took a prior incarnation of this class)
- A few examples of proteins with internally repetitive sequences: 1, 2, 3
Jan 19, 2023 - Sequence Alignment I
- Science news of the day, relevant to our discussion of ChatGPT last class: CNET & other news sources used it to write articles; this Gizmodo story reports that "the AI-program fabricates information and bungles facts like nobody’s business" and CNET "has been forced to issue multiple, major corrections". So, if you do opt to try ChatGPT to help with Python, be sure to check (and then double-check) everything.
- Today's slides
Problem Set I, due 10PM Jan. 30, 2023:
- Problem Set 1
- H. influenzae genome. Haemophilus influenza was the first free living organism to have its genome sequenced. NOTE: there are some additional characters in this file from ambiguous sequence calls. For simplicity's sake, when calculating your nucleotide and dinucleotide frequencies, you can just ignore anything other than A, C, T, and G.
- T. aquaticus genome. Thermus aquaticus helped spawn the genomic revolution as the source of heat-stable Taq polymerase for PCR.
- 3 mystery genes (for Problem 5): MysteryGene1, MysteryGene2, MysteryGene3
- *** HEADS UP FOR THE PROBLEM SET *** If you try to use the Python string.count function to count dinucleotides, Python counts non-overlapping instances, not overlapping instances. So, AAAA is counted as 2, not 3, dinucleotides. You want overlapping dinucleotides instead, so will have to try something else, such as the python string[counter:counter+2] command, as explained in the Rosalind homework assignment on strings.
Extra reading, if you're curious:
- BLOSUM primer
- The original BLOSUM paper (hot off the presses from 1992!)
- BLOSUM miscalculations improve performance
- There is a good discussion of the alignment algorithms and different scoring schemes here
Jan 17, 2023 - Intro to Python II
- Reminder that today will be part 2 of the "Python boot camp" for those of you with little to no previous Python coding experience. We'll be finishing the slides from last time, plus Rosalind help & programming Q/A.
- *** Rosalind assignments are due by 10 PM January 18. ***
- We'll talk a bit about ChatGPT today for co-programming
- Another strong recommendation (really) to the Python newbies to download Eric Matthes's GREAT, free Python command cheat sheets that he provides to accompany his Python Crash Course book.
Jan 12, 2023 - Intro to Python
- STANDARD REMINDER: My email inbox is always fairly backlogged (e.g., my median time between non-spam emails was 11 minutes when I measured it some time ago, and it's gotten much worse since then), so please copy the TA on all emails to help us make sure they get taken care of.
- Today's slides. Remember that today and the next lecture are dedicated to the Python Boot Camp to start getting those of you with minimal coding skills up to speed on the basics. Advanced programmers can skip class!
- E. coli genome
- Don't forget that the Rosalind assignments are due by 10 PM January 18. Please do start if you haven't already, or you won't have time to get help if you have any issues installing Python.
- Python 2 vs 3? Bioinformatics researchers held out for 2 until quite recently, but the shift to 3 is pretty clear now. We'll use Python 3 (the latest version in Anaconda is 3.9, but any recent version will be fine), but Rosalind and some materials are only available in Python 2.7, so we'll generally try to be version agnostic for compatibility. Use whichever you wish, but be aware that support for Python 2.7 has officially been stopped. For beginners, the differences are quite minimal and are summarized in a table here. There's also a great cheat sheet here for writing code compatible with both versions.
Jan 10, 2023 - Introduction
- Today's slides
- We'll be conducting homework using the online environment Rosalind. Go ahead and register on the site, and enroll specifically for BCH394P/364C (Spring 2023) Systems Biology/Bioinformatics using this link. Homework #1 (worth 10% of your final course grade) has already been assigned on Rosalind and is due by 10:00PM January 18.
Here are some online Python resources that you might find useful:
- First and foremost, and very, very useful if you're a complete Python newbie: Eric Matthes's Python Crash Course book. He made some GREAT, free Python command cheat sheets to support the book.
- Practical Python, worth checking out!
- If you have any basic experience at all in other programming languages, Google offered an extremely good, 2 day intro course to Python (albeit version 2) that is now available on Youtube.
- Khan Academy has archived their older intro videos on Python here (again, version 2)
Syllabus & course outline
An introduction to systems biology and bioinformatics, emphasizing quantitative analysis of high-throughput biological data, and covering typical data, data analysis, and computer algorithms. Topics will include introductory probability and statistics, basics of Python programming, protein and nucleic acid sequence analysis, genome sequencing and assembly, proteomics, synthetic biology, analysis of large-scale gene expression data, data clustering, biological pattern recognition, and gene and protein networks.
Open to graduate students and upper division undergrads (with permission) in natural sciences and engineering.
Prerequisites: Basic familiarity with molecular biology, statistics & computing, but realistically, it is expected that students will have extremely varied backgrounds. Undergraduates have additional prerequisites, as listed in the catalog.
Note that this is not a course on practical sequence analysis or using web-based tools. Although we will use a number of these to help illustrate points, the focus of the course will be on learning the underlying algorithms, exploratory data analyses, and their applications, esp. in high-throughput biology. By the end of the course, students will know the fundamentals of important algorithms in bioinformatics and systems biology, will be able to design and implement computational studies in biology, and will have performed an element of original computational biology research.
Most of the lectures will be from research articles and slides posted online, with some material from the...
Optional text (for sequence analysis): Biological sequence analysis, by R. Durbin, S. Eddy, A. Krogh, G. Mitchison (Cambridge University Press),
For biologists rusty on their stats, The Cartoon Guide to Statistics (Gonick/Smith) is very good. A reasonable online resource for beginners is Statistics Done Wrong.
Some online references:
An online bioinformatics course
Online probability & stats texts: #1, #2 (which has some lovely visualizations)
No exams will be given. Grades will be based on online homework (counting 30% of the grade), 3 problem sets (given every 2-3 weeks and counting 15% each towards the final grade) and an independent course project (25% of final grade), which can be collaborative (1-3 students/project). The course project will consist of a research project on a bioinformatics topic chosen by the student (with approval by the instructor) containing an element of independent computational biology research (e.g. calculation, programming, database analysis, etc.). This will be turned in as a link to a web page. The final project is due by 10 PM, April 12, 2023. The last 3 classes will be spent presenting your projects to each other. (The presentation will account for 5/25 points of the project grade.)
If at some point, we have to go into coronavirus lockdown, that portion of the class will be web-based. We will hold lectures by Zoom during the normally scheduled class time. Log in to the UT Canvas class page for the link, or, if you are auditing, email the TA and we will send the link by return email. Slides will be posted before class so you can follow along with the material. We'll record the lectures & post the recordings afterward on Canvas so any of you who might be in other time zones or otherwise be unable to make class will have the opportunity to watch them. Note that the recordings will only be available on Canvas and are reserved only for students in this class for educational purposes and are protected under FERPA. The recordings should not be shared outside the class in any form. Violation of this restriction could lead to Student Misconduct proceedings.
Online homework will be assigned and evaluated using the free bioinformatics web resource Rosalind.
All projects and homework will be turned in electronically and time-stamped. No makeup work will be given. Instead, all students have 5 days of free “late time” (for the entire semester, NOT per project, and counting weekends/holidays). For projects turned in late, days will be deducted from the 5 day total (or what remains of it) by the number of days late (in 1 day increments, rounding up, i.e. 10 minutes late = 1 day deducted). Once the full 5 days have been used up, assignments will be penalized 10 percent per day late (rounding up), i.e., a 50 point assignment turned in 1.5 days late would be penalized 20%, or 10 points.
Homework, problem sets, and the project total to a possible 100 points. There will be no curving of grades, nor will grades be rounded up. We’ll use the plus/minus grading system, so: A= 92 and above, A-=90 to 91.99, etc. Just for clarity's sake, here are the cutoffs for the grades: 92% = A, 90% = A- < 92%, 88% = B+ < 90%, 82% = B < 88%, 80% = B- < 82%, 78% = C+ < 80%, 72% = C < 78%, 70% = C- < 72%, 68% = D+ < 70%, 62% = D < 68%, 60% = D- < 62%, F < 60%.
Students are welcome to discuss ideas and problems with each other, but all programs, Rosalind homework, problem sets, and written solutions should be performed independently (except the final collaborative project). Students are expected to follow the UT honor code. Cheating, plagiarism, copying, & reuse of prior homework, projects, or programs from CourseHero, Github, or any other sources are all strictly forbidden and constitute breaches of academic integrity and cause for dismissal with a failing grade, possibly expulsion (UT's academic integrity policy). In particular, no materials used in this class, including, but not limited to, lecture hand-outs, videos, assessments (papers, projects, homework assignments), in-class materials, review sheets, and additional problem sets, may be shared online or with anyone outside of the class unless you have the instructor’s explicit, written permission. Any materials found online (e.g. in CourseHero) that are associated with you, or any suspected unauthorized sharing of materials, will be reported to Student Conduct and Academic Integrity in the Office of the Dean of Students. These reports can result in sanctions, including failure in the course.
The final project web site is due by 10 PM April 12, 2023.
- How to make a web site for the final project
- Google Site: https://sites.google.com/new