1. Bioinformatics training programs at the UdeM
Gertraud Burger
Robert Cedergren Center for Bioinformatics and Genomics
Biochemistry Department
Université de Montréal
Interdisciplinary Science Workshop
University of Manitoba, Winnipeg, Ma
15-16 Nov 2018

2. Overview Bioinformatics is …
History of program development and implementation
Underlying training philosophy
Program design considerations
Implementation issues
Maintenance & viability

3. 1. What is bioinformatics
interface between numerical (computer science, mathematics) and life sciences (biology, biochemistry, microbiology, ecology);
conceives analytical strategies & algorithms, analyzes data from ‘Omics, biomolecules, organismal & molecular interactions, evolution...

4. 2. History Bioinformatics training programs at all levels
BSc, MSc, PhD
Design period
Designed by
Dept Biochem (GB, Aubry); Dept Computer Sci (El-Mabrouk, Boyer, L’Ecuyer, Major)
Implemented by
GB, El-Mabrouk
Programs offered since
BSc: Fall 2001; MSc, PhD: Fall 2002
Program modifications
~all 4 years
My role
design, implementation, president of study committees for 10 y.

5. 3. Training scope Agree on training scope:
broad = encompassing a max. of bioinfo. fields (more superficial)
deep = focus on basics (more fundamental)
specialized
training for job market or graduate studies?
We agreed on more fundamental training, and mostly for graduate studies.
specialized
broad
Imag anal
syst biol
phylogeny
ont, DBs
machlearn
Omic
mol struc
math
phys
chem
biol
comp S
deep

6. 3. Training scope cnt’d 1
a. Issue: How much life-science hands-on for dry-lab bioinformaticians?
Data analysis requires an understanding of biological data
nature
accuracy
trustworthiness
To make aware of potential problems and pitfalls in data production, we included biochemistry wet-lab courses in program.
b. Problem with ‘deep scope’ (emphasis on fundamental courses)
relevance to/connection with bioinformatics not readily evident (as basic courses are shared by many different programs)  motivation loss
difficulty to learn simultaneously multiple distinct disciplines.

7. 3. Training scope cnt’d 2
Current program (after 10 y, three amendments, compromises, constraints…)
Courses
Credits
02A
02B
Fundamentals
math (calculus, algebra)
probability & statistics
4-8 cr
01A
Bioinformatics
Basics; tools; internship
16 cr
01B
Informatics
Programming, discrete structures, data structure
12 cr
01C
Life Science
biochemistry +TP, DNA-RNA-protein, genetics,
14 cr
02C
Optional Informatics
Optimization, algorithmics, DB, comp. graphics, software design
12-21 cr
02D
Optional Life science
metabolism, cell biology, organic chemistry, microbiology, ethics
8-18 cr
02Z
Specialization
Genetics/ molec.medicine/ comp.languages/ theor. comp.sci/
Statistics&machine Learning
0-12 cr
02Y
Free choice
English, evolution, biodiversity, management, …
0-6 cr
reality sets in
Cut back in fundamental courses: no physics (from the beginning)
chemistry not obligatory (recent); critical notions treated in biochemistry courses.

8. 4. Program design a. Format of program b. Program content
influenced by training scope
we chose regular 90-credit format (3-y BSc; shorter programs (‘micro’ program, ‘Certificat’) only suited for specialized, narrow training)
b. Program content
proportion & choice of various traditional disciplines
mathematics, physics, chemistry, biology, computer science
algebra, calculus, statistics, probabilistic; mechanic, thermodynamics, optics; organic, inorganic, physical chemistry; microbiology, structural biochemistry, molecular biology, genetics, physiology, enzymatic, evolutionary biology; programming, discrete structures, data structure, algorithmic, databases, machine learning, …
theoretical vs practical courses & research internships
proportion & choice of complementary courses
ethics, communication, philosophy, management; free choice.
proportion of integrative (true bioinformatics) courses
important to keep sight of program’s goal

9. 4. Program design ctn’d Problems we encountered
how to fit all relevant courses in a 90 cr. envelop
how to establish a conflict-free course schedule (given shared basic courses)
account for course prerequisites / succession of courses
high workload, esp. 1st year
offer each trimester at least one integrative course.

10. 5. Implementation issues
Admission criteria
R score
type of Diploma of Collegial Studies, e.g. in Science, Lettres, Art, etc)
b. Budget
program management by pre-existing department w/o separate budget or own dept with dedicated budget?
autonomy through strategic training grant (CIHR, 310 k/y)
trans-disciplinarity involves of multiple departments, faculties with different degrees of commitments
share infrastructure, lab instructors
share scholarships, study fee reductions
invest in publicity
teaching load in other programs recognized by departments?
c. Max nr. of students/y; ‘paying customers’ vs training quality

11. 6. Maintenance & viability
a. Study committee must be composed of teachers from different disciplines
b. Stable, foreseeable institutional support required (deployed teachers, budget, admin support).
c. Continuous program follow-up: nr. admissions, perseverance, students’ background, performance/course, path (after abandoning, graduation), career opportunities  to find suited measures for improvements
d. Continuous course content analysis
alignment across courses to avoid overlaps and omissions
integration of new contents.

12. 6. Maintenance & viability cont’d
Problems
a. Universities stress large nr. of admissions & throughput
nr. of admissions & throughput affect training/trainee quality
nr. of applications depends on awareness about & visibility of program
b. Nr. of bioinformatics graduates is below average, due to
demanding program esp. in 1st year
difficulty to perform well in multiple disciplines
performance comparison with mono-disciplinary trainees
preceding training may not be fully suited
requires adjustments
extra tutors
reduction of courses in 1st year
basic courses (math, stat, programming, biochem) tailored for bioionformatics.