Self study genomics and data
Week 1
The 12 principles for data organization in spreadsheets
1. Be consistent: organize and enter the data consistent from the start. Use consistent
codes for categorical variables and for any missing values. Also use consistent variable
names, subject identifiers, file names and data layout.
2. Choose good names for things: do not use spaces because it make programming
harder. Use ‘’_’’ instead of spaces or write it without any spaces. Do not use mixtures
and avoid special symbols. The name has to be short but meaningful.
3. Write data’s as YYYY-MM-DD: other ways excel can cause problems in the data.
4. No empty cells: fill in “NA” in the missing data cells to prevent empty cells.
5. Put just one thing in a cell: every cell should contain just one peace of data. It is
better to write “body_weight” instead of “body_weight_g” and “45” instead of 45_g.
6. Make it a rectangle: try to make the data in a rectangle or a set of rectangles
7. Create a data dictionary: folder explaining the data and telling other important
information about the dataset
8. No calculations in the raw data: make calculations in a copy of the data
9. Do not use colour to convey information: create a separate column expressing what
the colour was supposed to express
10. Make back ups: back up your data multiple times in multiple locations so if the data is
corrupted, there is always a place to go back
11. Use validation to avoid errors: use validations mechanisms in excel to prevent errors
when typing in the data
12. Save data in plain text files: better converted into the csv files
Five most common problems with messy data
Column headers are values, not variable names
Multiple variables are stored in one column
Variables are stored in both rows and columns
Multiple types of observational units are stored in the same table
A single observational unit is stored in multiple tables
Properties of tidy data
Each variable forms a column
Each observation forms a row
Each type of observational unit forms a table
, Week 2
P- hacking: The inappropriate manipulation of data analysis to enable
a favoured result to be presented as statistically significant
Open science is making research results available for anyone
FAIR-principles
o Findable online available resources
o Accessible access to the metadata (information about the data)
o Interoperable understandable for others
o Reusable data in community standards to be reusable
Data is not really fair or unfair, it is more of a spectrum
Motivations for open science
o Reproducibility: a published data analysis is reproducible if the analytic
datasets and the computer code used to create the data analysis is made
available to others for independent study and analysis
o Replicability: the independent investigation of a scientific hypothesis with
newly collected data, new experimental setups, new investigators, and
possibly new analytic approaches. This approach allows for differences in
results that arise from statistical variability
o Repeatability: this is the possibility to perform the experiment multiple times
under the same conditions, describes variation in successive measurements
Motivations against open science
o Privacy concerns
o Difficult to implement, expensice and time-consuming
o Ownership of data is important
o Concerns of bad-faith actors
Engaging in open science: reliable, reproducible, reusable, and relevant
Pillars of open science: open data, access, methodology, source, peer review,
education
The reproducibility of research will bring:
o Learn for others about how best to analyse certain data
o Reduce human errors as data become larger and more complex
o Free up time for re-analysers to focus on parts of a data analysis that require
more human interpretation.
o have discussions about what makes for a good data analysis in certain areas
o Improve the quality of future data analyses
Week 1
The 12 principles for data organization in spreadsheets
1. Be consistent: organize and enter the data consistent from the start. Use consistent
codes for categorical variables and for any missing values. Also use consistent variable
names, subject identifiers, file names and data layout.
2. Choose good names for things: do not use spaces because it make programming
harder. Use ‘’_’’ instead of spaces or write it without any spaces. Do not use mixtures
and avoid special symbols. The name has to be short but meaningful.
3. Write data’s as YYYY-MM-DD: other ways excel can cause problems in the data.
4. No empty cells: fill in “NA” in the missing data cells to prevent empty cells.
5. Put just one thing in a cell: every cell should contain just one peace of data. It is
better to write “body_weight” instead of “body_weight_g” and “45” instead of 45_g.
6. Make it a rectangle: try to make the data in a rectangle or a set of rectangles
7. Create a data dictionary: folder explaining the data and telling other important
information about the dataset
8. No calculations in the raw data: make calculations in a copy of the data
9. Do not use colour to convey information: create a separate column expressing what
the colour was supposed to express
10. Make back ups: back up your data multiple times in multiple locations so if the data is
corrupted, there is always a place to go back
11. Use validation to avoid errors: use validations mechanisms in excel to prevent errors
when typing in the data
12. Save data in plain text files: better converted into the csv files
Five most common problems with messy data
Column headers are values, not variable names
Multiple variables are stored in one column
Variables are stored in both rows and columns
Multiple types of observational units are stored in the same table
A single observational unit is stored in multiple tables
Properties of tidy data
Each variable forms a column
Each observation forms a row
Each type of observational unit forms a table
, Week 2
P- hacking: The inappropriate manipulation of data analysis to enable
a favoured result to be presented as statistically significant
Open science is making research results available for anyone
FAIR-principles
o Findable online available resources
o Accessible access to the metadata (information about the data)
o Interoperable understandable for others
o Reusable data in community standards to be reusable
Data is not really fair or unfair, it is more of a spectrum
Motivations for open science
o Reproducibility: a published data analysis is reproducible if the analytic
datasets and the computer code used to create the data analysis is made
available to others for independent study and analysis
o Replicability: the independent investigation of a scientific hypothesis with
newly collected data, new experimental setups, new investigators, and
possibly new analytic approaches. This approach allows for differences in
results that arise from statistical variability
o Repeatability: this is the possibility to perform the experiment multiple times
under the same conditions, describes variation in successive measurements
Motivations against open science
o Privacy concerns
o Difficult to implement, expensice and time-consuming
o Ownership of data is important
o Concerns of bad-faith actors
Engaging in open science: reliable, reproducible, reusable, and relevant
Pillars of open science: open data, access, methodology, source, peer review,
education
The reproducibility of research will bring:
o Learn for others about how best to analyse certain data
o Reduce human errors as data become larger and more complex
o Free up time for re-analysers to focus on parts of a data analysis that require
more human interpretation.
o have discussions about what makes for a good data analysis in certain areas
o Improve the quality of future data analyses
