High-throughput, high-precision colony phenotyping with pyphe

doi:10.21203/rs.3.rs-401914/v1

Download PDF

Method Article

High-throughput, high-precision colony phenotyping with pyphe

https://doi.org/10.21203/rs.3.rs-401914/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Colony fitness screens are powerful approaches for functional genomics and genetics. This protocol describes experimental and computational procedures for assaying the fitness of thousands of microbial strains in numerous conditions in parallel. Data analysis is based on pyphe, an all-in-one bioinformatics toolbox for scanning, image analysis, data normalisation and interpretation. We describe a standard protocol where endpoint colony areas are used as fitness proxy and two variations on this, one using colony growth curves and one using colony viability staining with phloxine B. Different strategies for experimental design, normalisation techniques and quality control are discussed. Using these approaches, it is possible to collect hundreds of thousands of data points, with low technical noise levels around 5%, in an experiment typically lasting two weeks or less.

General Microbiology

Screen

Colony

Fitness

Functional genomics

Phenomics

Microbiology

Python software

large-scale phenotyping

Cell viability

Growth curve

Measuring properties of densely arrayed microbial colonies is a powerful and efficient technique for determining the fitness of large panels of microbial strains. Today, numerous knock-out or over-expression mutant libraries, wild strain collections, segregant panels and synthetic genetic array construction methods are available to underpin systematic, data-driven investigations. Owing to specialised robotics, colony screens are largely automated and can be done at huge scales, e.g. with millions of double knock-out mutants (1). Colony screens have been employed to measure the fitness of large segregant libraries for the dissection of complex traits (2, 3) and the characterisation of wild strain libraries (4–6). All such experiments require scalable computational approaches for data acquisition, data management, batch correction and statistical analysis. To this end, we have recently published a new toolbox named pyphe which enables fast, reproducible and flexible pipelines for colony screen analysis (Figure 1) (7).

One key challenge of colony screens is that the observed colony phenotype is affected by factors other than the strain’s fitness. Batch effects, e.g. originating from media preparation or incubation time, make colony sizes not directly comparable between different plates. Even within plates, nutrient, moisture and temperature gradients as well as differences in inoculation mass due to uneven pinning can result in variation between colonies located in different areas of the plate. A well-known example is the “edge effect”, where outermost colonies grow larger due to a lower degree of nutrient competition. Normalisation is therefore key for obtaining reliable fitness estimates. Edge effects, and other effects that affect rows and columns evenly, can in some cases be rectified by dividing by the row/column median, and other, regional biases can sometimes be rectified by comparing colonies to their neighbours (e.g. as implemented in SGAtools (8)). Both these normalisation methods only work if the null effect can be reliably estimated, i.e. if the majority of colonies show no growth effect. An alternative approach is to include a grid of control strains on every plate (Figure 2) (9). Each colony is then compared to essentially a weighted average of the neighbouring controls, resulting in a relative fitness that is standardized within and across plates (as long as the same control strain is used). Pyphe implements row/column median normalisation as well as a grid normalisation, giving the user full control over the normalisation strategy. Using grid normalisation, it is possible to achieve technical noise levels of around 5%, enabling the detection of subtle differences in fitness.

Classically, plates are imaged once and colony footprint areas are used as fitness proxy. Recent methods have expanded on this in two areas. ScanLag (10), Quantitative Fitness Analysis (11), Colony-live (12) and Scan-o-matic (9) use repeated imaging of colonies and extract parameters from the resulting growth curve. Additionally, Colony-live and Scan-o-matic use transmission scanning where the grey-value of a pixel can be used as a proxy for colony thickness. These methods achieve lower noise values than endpoint measurements but require substantially higher investments in scanners (which need to be housed in a temperature-controlled environment), data storage and analysis. We recently found that, at least within of the parameters tested, the growth rate parameters extracted from timeseries correlate tightly with endpoint sizes (7). This is compatible with a model where colonies grow until the media is exhausted by the ensemble of colonies, giving each colony a fixed time-window for growth. This is an important difference to screens conducted in multi-well plates with liquid media, where each culture is given a fixed amount of resources. Pyphe supports the acquisition and analysis of colony growth curves as well as traditional endpoint measurements, allowing the user to choose the method most suitable for their experiment and laboratory setup.

Another variation on classical screens is the use of phloxine B as a dead cell stain (13). If this is included in the media, the colour of a colony reflects the proportion of dead cells in it (7). This provides an orthogonal readout that is often uncorrelated to colony sizes after both readouts are standardised to the control strain. As phloxine B in the media generally does not affect cell growth, it can be included in all plates, whereby a second fitness dimension is added to the experiment at little extra cost and effort. Pyphe can acquire and analyse colour images and extract quantitative redness scores which then integrate seamlessly into the rest of the analysis pipeline.

We here present a detailed protocol for colony fitness screens, split into an experimental and computational part. The basic protocol describes an endpoint experiment using colony areas as fitness proxies. In our opinion, this should be the default for most investigations and the starting point for new users. Two variations on this are described: How to record and analyse colony growth curves and how to analyse colony viability by phloxine B staining. Numerous notes explain details and important considerations regarding experimental design and quality control.

Experimental

Sterile microbial growth media (liquid and with agar)
Rectangular agar plate dishes, e.g. Singer PlusPlates
Microwave (or autoclave) for melting agar media
50ml serological pipettes and pipette pump
Conical flask (e.g. 250ml), shaking incubator
Pinning robot and consumables (e.g. Singer RoToR with 96 long RePads, 96 short RePads and 1536 short RePads)
Cling film
Plate incubator
Strain library cryostocks in 96 or 384 format, ideally with unique footprints (see Note 2)
Cryostock of a suitable control strain

Optional

Incubator to house scanners (for timecourse imaging)
Phloxine B (for colony viability measurements)
Reagents required for testing specific growth conditions of interest (e.g. drugs, chemicals or nutrients)

Scanning and Data Processing

Scanner (Epson V800 or V850 recommended) with a fixture to hold plates into place on the scanner surface (laser cutting guide available on https://github.com/Bahler-Lab/pyphe/tree/master/Documentation/Scan). Please see Figure 3 for more information on how to set up the scanner.
A computer with Ubuntu (or another Linux operating system) and Python3 (e.g. Anaconda3) for scanning. To install pyphe, run the following command in your terminal
$ pip install pyphe
On this computer, ImageMagick needs to be installed, the scanner needs to be accessible with SANE (Scanner Access Now Easy) and the TPU8x10 scanning mode must be available. This is the case by default for recent versions of SANE, which is included in the standard Ubuntu installation. The pyphe manual page contains further information on how to set up the scanner. For older versions, scan-o-matic (9) has developed some custom solutions for setting up the scanner.
After scanning, the rest of the analysis can be performed on a computer with any operating system with Python3 and pyphe To use pyphe, you need to be familiar with some standard characteristics of command line programs, especially the concept of the working directory and relative file paths. In the terminal, navigate to the base directory of your experiment and run pyphe commands from there. File paths need to be defined in relation to that directory. Commands in this protocol are marked with $, indicating that this is a line of code to be run in the terminal (the $ is not part of the command). Help for each pyphe tool is available by running the tool with the --help parameter (e.g. $ pyphe-analyse --help)

General Method: Endpoints

Experimental

Plan and prepare your experiment

Pick an appropriate control strain for your library. For knock-out, over-expression and similarly constructed libraries this should usually be the library background strain (“wild type”). For collections of wild isolates or segregant libraries, the choice of control strain is more flexible, but it should not fall on either extreme side of the distribution of fitness phenotypes.
Plan the arrangement of library and control strains on the final assay plates. Overall, five different types of plates are required throughout the workflow: library plates, grid plates, combined plates, source plates and assay plates (Figure 2). Consider how many replicates are required to answer your biological question (Note 1). Plan how to rearrange the library plates around the control grid and prepare the layout file of your assay plates (Notes 2 and 3).
Prepare sterile growth media and pour agar plates for waking up the library and for the preparation of the grid plate (Note 4). We usually use rich media for waking up strains and preparing the grid plate, even if the screen is in the end carried out on minimal media.
Wake up the control strain and your library. This protocol is applicable to a wide range of microbial organisms and growth media, so we cannot specify incubation times for plates. The general advice for the preparatory phase is that plates should be incubated until colonies are fully (or almost fully) grown as this makes pinning easier, however, not so long that colonies get stressed due to starvation or desiccation.

Assemble the combined plates and make source plates

Prepare grid plates by growing a 40ml liquid culture of the control strain overnight. Then, pour the culture into an empty PlusPlate and, using 96 long RePads, pin from liquid to solid to make the grid plates. Do not use target mixing. Prepare multiple grid plates as required (plus some spares) as the number of times a grid plate can be pinned from is limited to approximately 10.
Once grown, re-array grid and library plates as shown in Figure 2. Use 96 short RePads for this. On the Singer RoToR, you can either use the inbuilt 16-to-1 array mode or make your own pinning program in custom mode.
To make the source plates, copy the combined plates onto fresh agar plates, using 1536 short pins. This step is important because the combined plates are a patchwork and should therefore be copied again to make consistent, evenly arrayed source plates. Besides, this step can serve as an amplification step for larger screens with many assay plates. Simply prepare several copies of each source plate, approximately one for every 10 assay plates.

Make assay plates

Prepare your assay plates with drugs, inhibitors or particular nutrient composition as required (Note 4). Make one or two spares per condition in case something goes wrong during pinning.
Copy your source plates onto the assay plates using 1536 short pins. Consistent, small inoculums are obtained with low pinning pressure (5-10%). Make sure the pinning works and does not miss any corners (Note 6). Clearly label each plate with a unique ID (Note 5).
Wrap plates in cling film and place in an incubator. Plates should always rest upside-down to prevent condensate dropping onto the colonies. Ideally, there should be no strong air currents or temperature gradients in the incubator which can lead to plates drying out and colonies growing unevenly.
Let the colonies grow until they stop or barely grow anymore (usually 2-3 days). Please see (7) for detailed reasoning for this choice of time point.

Analysis

1. Acquisition.

Image your plates once they have grown. Turn on the scanner and run
$ pyphe-scan --fixture som3 --postfix POSTFIX --nplates N
Where POSTFIX is the name of your experiment and N the total number of plates you have to scan. The fixture argument specifies how to crop individual plate images from the original scan. Parameters for our fixture model are pre-programmed (use som3_edge and som3 for images containing or not containing the edge respectively), but it is possible to configure your own fixtures in the pyphe-scan script. Plates are always placed in the scanner with the agar on the bottom (i.e. not upside-down) and with the A1 colony position in the back-right corner (see Figure 3). Remove the plate lids, lower the scanner top and make sure it is even. Press ‘y’ when prompted to start scanning. Swap plates as instructed. Pyphe-scan will create a folder with the current date and the POSTFIX in the current directory in which images will be saved.
In parallel, prepare a table mapping the scan sequence to the plate ID. Later, expand this to include information about the layout, experimental condition and any additional meta-data (date, batch, comments etc.) you want to keep track of. This will be your experimental design table (EDT, see Note 5).

2. Quantification.

a. Extract colony sizes from images by running

$ pyphe-quantify batch --grid auto_1536 --pattern “images/*.jpg”
The batch parameter indicates the mode in which to run pyphe-quantify. In batch mode images are analysed one at a time and colony areas are extracted. The --grid option specifies how to assign identified colonies to grid positions. In automatic mode, as used here, the grid is detected based on pixel row and column intensity peaks. There are other ways to manually define grid positions when this is required, please see the documentation. The --pattern argument can be used to specify which images should be analysed, in this case, it is assumed that all your images are located in a folder ‘images’ in the current working directory and are in jpg format.

b. Carefully check whether the results are as expected. A QC image is produced for every single analysed image. Investigate if colonies were identified correctly and that colonies are assigned the correct row and column. Pyphe-quantify has numerous parameters that can be adjusted, check the documentation for details.

c. Edit your EDT table to include a column ‘Data_path’ which contains relative file paths to each image’s data file (these are located in the ‘pyphe_quant’ directory by default). If you have to exclude individual images for QC reasons, simply delete the corresponding lines from the EDT table.

3. Analysis

Aggregate data across plates and perform grid and subsequent row/column median normalisation by running
$ pyphe-analyse --edt EDT.csv --format pyphe-quantify-batch --gridnorm standard1536 --rcmedian --qc_plots analyse-qc
The --edt option specifies the relative path to your experimental design table and should be adjusted if required. The --format option describes the type of input data, in this case colony areas produced by pyphe-quantify. The --gridnorm argument enables grid correction and the following parameter specifies where the grid colonies are located on the plate. “standard1536” refers to the recommended position of two 96 grids in the top left and bottom right corners. In some cases, when your grid strain grows differently than your library, it might not be appropriate to use row/column normalisation (see Note 8) after grid correction. In those cases, simply leave out the --rcmedian parameter. The --qc_plots option tells pyphe-analyse to produce a QC report for each plate in the experiment, which will be saved in the specified folder. There are additional options which are explained in the documentation.
Once the command has finished, join your layout table onto the produced colony report (in R or python or the programming language of your choice). Alternatively, you can use the --load-layouts option to load strain IDs directly into your colony report.
Perform quality control suited to your experiment (Note 9). For each plate, check that the footprints are empty and check the coefficient of variation of the additional grid colonies. Discard plates from the analysis as required.

4. Interpretation

Compute summary statistics and p-values comparing each strain to the control strain (repeated across all conditions).
$ pyphe-interpret --ld pyphe-analyse_data_report.csv --circularity 0.85 --control CONTROL-STRAIN --grouping_column Condition --axis_column Strain

The --ld option specifies the path to the colony report generated in the previous step. The --circularity option triggers a filter that excludes all colonies with a circularity below the specified value. This is useful in order to exclude image analysis artefacts from further analysis. The --grouping_column argument specifies a column in the colony report for subsetting your data. In each subset, i.e. for each unique entry in the grouping column, a different set of t-tests will be performed. The --axis_column argument specifies the explanatory variable of your t-test. The --control option specifies your control strain and this needs to be the string identifier also found in the axis column of your colony report.

and/or

b. Compute summary statistics and p-values comparing each condition to the control condition (repeated across all strains). This tests for condition-specific differences in fitness, correcting for basal growth of the strain.
$ pyphe-interpret --ld pyphe-analyse_data_report.csv --circularity 0.85 --control CONTROL-CONDITION --grouping_column Strain --axis_column Condition
For example, in a scenario where a slow-growing mutant (relative to the wild-type grid strain) has been measured in a large number of conditions and shows a consistent growth defect (e.g. a fitness of 0.8) in all conditions (including a control condition without stress), the first command would result in a significant p-value for this strain in all conditions as each time it is compared only to the wild type (which has a fitness of 1) in the same condition. This second approach would show no significant difference in fitness as the fitness in each condition is the same as in the control condition.

c. Proceed to analyse your data according to your question. For calling hits, it is usually useful to define an effect size threshold additional to your p-value cut-off.

Variation 1: Time course analysis

Pyphe implements methods to analyse growth curves of colonies, obtained by automated, regular imaging of plates. Characteristics of the growth curve, usually the maximum slope, are extracted and corrected for plate position effects, similarly as endpoint colony sizes. This method typically achieves lower noise values, but we nonetheless only recommend it for specialised applications as the maximum slopes extracted from growth curves correlates tightly with much more easily obtained endpoints (7).

To record and analyse growth curves, prepare your assay plates as described in the Experimental section above. Then proceed as follows:

Immediately after pinning, place the assay plates into the scanner without lids. Take a note of the plate ID corresponding to each position. Close the lid so that it rests horizontally on the plates. If desired, place the scanner in an incubator or temperature-controlled environment.
On the scanner computer, run pyphe-scan-timecourse.
$ pyphe-scan-timecourse --nscans NSCANS --interval INTERVAL --postfix POSTFIX --fixture som3
NSCANS is the number of scans in the timecourse, INTERVAL is the time between the end of a scan and the start of the next (we recommend 20 to 30 minutes). After the first scan, check that everything is working as expected and that the first images look ok.
After scanning has finished, manually inspect the images. Often, it is necessary to remove some images at the end of the timecourse, e.g. because colonies started touching each other, contaminations or condensation appeared, or the plates dried out.
For every plate separately, extract growth curves using pyphe-quantify in timecourse mode.
$ pyphe-quantify timecourse --grid auto_1536
This will produce a table of growth curves, where individual colonies are the columns and timepoints are the rows.
For every plate separately, extract growth curve parameters using pyphe-growthcurves.
$ pyphe-growthcurves --input INPUT --plots
Input is the path of the csv file produced by pyphe-quantify.
Proceed with analysis as above. Run pyphe-analyse using the --format pyphe-growthcurves option. The Data_path column of your EDT table should contain paths to the files produced by pyphe-growthcurves. Pyphe-analyse will perform corrections on the maximum slope of the growth curves but will retain and include other growth curve parameters in the colony report.

Variation 2: Colony viability analysis

Addition of the dead-cell stain phloxine B to the media results in colonies with varying redness which is correlated to the fraction of dead cells in the colony. This can be performed in parallel to the standard workflow, imaging each plate twice: once in greyscale transmission mode (for colony sizes) and once in colour reflective mode (for viability analysis). In our hands, viability analysis with phloxine B works better in 384 format and in conditions which are not too stressful, which both results in larger colonies with a stronger signal.

Proceed to prepare your assay plates as above but include 5 mg/L phloxine B in the media. A 1000x aqueous stock can be stored in the fridge protected from light for several weeks.
Grow plates as usual, but protected from light (phloxine B is light sensitive)
Scan plates with the white background attached to the scanner lid.
$ pyphe-scan --mode Color --fixture som3-color --postfix POSTFIX --nplates N
Analyse the images using pyphe-quantify in redness mode
$ pyphe-quantify redness --grid auto_1536 --pattern “images/*.jpg”
Proceed with the analysis in pyphe-analyse. We recommend using only row/column median correction if possible.

$ pyphe-analyse --edt EDT.csv --format pyphe-quantify-redness --rcmedian --qc_plots analyse-qc

With high-density colony arrays, pinning robots and streamlined computational pipelines, it is possible to gather and analyse hundreds of thousands of fitness data points in an experiment lasting typically less than two weeks. Pyphe provides several command-line tools facilitating all steps of colony fitness screens, from image acquisition to data interpretation. This modular approach has two key advantages: First, it allows for flexibility by enabling the use of different fitness proxies as input and by allowing for the integration of other, external tools into the pipeline. Second, it makes data analysis reproducible and scalable by allowing them to be scripted.

We have described three possible protocols, using colony endpoints, colony growth curves and redness scores for the investigation of microbial fitness. These, together with the accompanying notes, are meant as a starting point for users to implement their own workflows. By providing a framework of data formats and analysis steps, much of it will be applicable to researchers not using pyphe but looking to implement their own computational or experimental procedures. Experimental details will vary between microbial species, assay conditions and laboratory settings. In particular phloxine B staining has only been used in a quantitative screening context in S. pombe, so some initial validation will be required when applied to other species. Pyphe is set up as an open-source, collaborative software project. We welcome user feedback and contributions for its continuous improvement.

Scan-o-matic: Scan-o-matic (9) is a sophisticated platform for scanning, image analysis and spatial normalisation. Scan-o-matic pioneered spatial correction with reference grids and additionally supports calibrating pixel intensities to cell numbers, enabling exact quantification of population sizes. Scan-o-matic uses timecourse imaging and requires a specialised hardware set-up, including pixel calibration strips, modifications to scanners and a local area network.

Gitter: Gitter (14) is an R package for determining colony sizes from images. It can work with a range of input image types and has robust algorithms for thresholding and grid assignment. Pyphe can work with image quantification data from gitter. Unfortunately, gitter is currently archived by CRAN as it requires outdated packages. For experts, it is possible to install it from the archive but this requires manual installation of several dependencies.

SGAtools: SGAtools (8) provides tools for spatial normalisation (not including reference grid correction) and statistical analysis of colony screen data. Available as a web service without installation.

Colonyzer and QFA: Colonyzer (15) has been developed for the Quantitative Fitness analysis workflow (11). Colonyzer has been designed to work on colonies which have been stamped on plates with manual replicators, resulting in a spot (i.e. a spread-out inoculum).

IRIS: IRIS (16) is an advanced image analysis tool for single or timecourse images, specialising in detecting additional colony morphology features such as bacterial biofilm formation.

CellProfiler: CellProfiler (17) is a powerful, multi-purpose image analysis tool with which one can assemble analysis pipelines to count and measure colony sizes as well as potentially other morphological parameters.

Note 1: Experimental design: power calculations

At the beginning of the planning phase of an experiment, consider the resolving power required of it to answer your biological question. This will allow you to design cost- and labour-efficient experiments. In the simplest case, you might only be interested in the binary classification of growth/no growth, e.g. when testing the essentiality of genes in specific contexts. In another setting, the goal might be to characterise growth differences down to a couple of percentage points, e.g. when identifying genetic variants of wild strains which have subtle effects on growth. The number of replicates required to address these two questions differs drastically. You can design your experiment appropriately by calculating the number of required replicates based on the minimal effect size you are trying to detect, the noise of the method (5-10% is a good, conservative estimate in general) and the desired statistical power (chance of correctly rejecting the null hypothesis). This can be done using the stats.power module of the python statsmodels package or the power.t.test function in R, or other online tools. For example, to have an 80% chance of detecting a 5% difference in growth at 5% noise (standardised effect size = 1), with a p-value cutoff of 0.05, you would require 17 samples in each group (two-sided Student's t-test). You also need to consider that correcting for multiple testing will in the end decrease the statistical power of your test. This effect is of course stronger the larger your strain library is. In the case of the Bonferroni correction (which is not recommended), you can easily calculate this loss of power by dividing your alpha by the number of tests. For preferred FDR methods, this is not straightforward and requires you to make an educated guess.

Note 2: Experimental design: plate layouts

T-tests generally require the samples to be independent. What this means in practice is often not entirely clear, and it will be up to you to decide the details of your experiment design. In general, we would not consider replicates located next to each other on the same plate as independent as they are subject to the same pinning errors as well as local nutrient, moisture and temperature regimes. It is therefore not recommended to generate replicates using 1-to-4 or 1-to-16 multiplexing pinning programmes where a single 96 plate is replicated into 384 or 1536 respectively. Instead, replicates should be obtained by recombining library plates in different combinations into the assay plate, mixing up the extract location and neighbours.

Colonies grow in competition and lack of neighbours usually results in increased growth. This is commonly observed as an edge effect where plates on the borders grow bigger. For the same reason, colony screens are sensitive to empty areas of the plate, effectively creating internal edges. If library plates are only partly filled, we recommend filling empty spots with control strains or random strains. This is less important if empty positions are scattered and more so if entire corners, areas, rows or columns are empty.

Despite generally discouraging empty positions, we strongly recommend leaving one individual, unique position per library plate empty. Such footprints serve two important purposes. Firstly, they are negative control positions which should always be empty, if they are not this indicates a source of contamination. Secondly, they aid identification of specific plates in the case of multiple assay plate layouts. Images will not contain the labelling information written on the side of the plates, so in case of a mix-up, the footprints can provide crucial clues to the identity of the plate.

Note 3: Preparation of layout files

Preparing the layout of your assay plates is a key task that will require some programming/data processing, at least for larger experiments. This is best done with layouts in long format, i.e. using a table with at least three columns (Row, Column, Strain) where every line of the table describes a single colony position. For every library plate in 96 format, take a note of the position into which it is pinned onto the combined plates. When pinning from 96 to 1536 format, there are 16 of these positions (rows 1-4 and columns 1-4). Let’s call the row position p_r and the column position p_c. The row and column position of a colony on the assay plate (1≤a_r≤32 and 1≤a_c≤48) is then related to the position (1≤s_r≤8 and 1≤s_c≤12) on the source plate by:

a_r = 4*(s_r-1) + p_r

a_c= 4*(s_c-1) + p_c

Using this formula, transform the row and column values for each 96 library plate depending on their target position. Create a layout for the grid plate and include it too. Then, concatenate the tables for all plates and sort by row and column.

Note 4: Preparation of suitable agar plates

To achieve high data completeness and low technical noise, it is crucial that assay plates are flat, without bubbles and of suitable dryness. For this, ensure the following:

Always let the media cool down to approximately 60°C before pouring plates. Otherwise, the contraction of the agar during cooling will result in unwanted ripples on the surface. When potentially temperature-sensitive drugs are to be included in the assay plates, add these when the media has cooled down and right before pouring to minimise the exposure to high temperatures.
Always pour plates on a level and even surface. In our experience, this can be done on a standard lab bench without a sterile environment, as long as the plate lid is immediately replaced after pouring.
Always add a consistent amount (we use 40ml) of agar medium to each plate. This will result in plates with a consistent height and also avoid other artefacts. Thicker plates mean more nutrients are available to each colony which will change the colony size. Use a serological 50ml pipette and take up 5ml more than required. This will prevent bubbles. If you spot any bubbles, suck them back up with the pipette.
Plates should be dried for a consistent time (we use 45 minutes without lids) before use. Alternatively, closed plates can be left unwrapped on the bench overnight which in our experience results in plates with a suitable dryness for immediate use. Plates can be stored in the fridge/cold room if not used immediately but will require extra drying before use.

Note 5: Preparation of the EDT table

Preparing a correct and complete experimental design table (EDT) is a key requisite for obtaining results with pyphe-analyse. All plates in your experiment should have a unique ID and this should be clearly written on the edge of the plate (not on the bottom where it would show up in the scanned images). For example, plate IDs could follow the format date_layout_condition_replicate. While scanning, take a note of which plate ID corresponds to which image in the scan sequence in table format. This table can then later easily be transformed into the EDT required by pyphe by adding extra columns with additional meta-data. The final table must be in csv format and the first column must contain the unique plate IDs. There must also be a column called ‘Data_path’ which points to the image data file produced by pyphe-quantify. Any additional meta-data, such as date, condition, library versions, comments, batch information, can be included and will be carried through to the colony report. Please see the Documentation folder in the pyphe github repository for an example file. Please note that there is no need for all the data files to be located in the same folder, which is convenient if you have large experiments containing several batches. Generally, file paths should not contain spaces or non-standard characters or those with special meanings in the terminal (%,>,?,/,”,*,&, etc.). Use _ or . or - to separate words.

Note 6: Quality control: pinning

Even with precise robots, the transfer of colonies by pinning can be prone to errors which, if unspotted, will result in missing or wrong data. A common problem that occurs when target plates are uneven is that entire areas or corners of the plate have no colony inoculums. Such pinning errors are dangerous as they could result in colonies which are absent for technical reasons to be interpreted as unviable phenotypes. Pyphe helps spot these errors by detecting missing control colonies and setting all neighbouring colonies to NA, but this only occurs after the damage is done. We therefore strongly advise to check every plate for missing corners and correct pinning by eye. Have one or two spare plates at hand to repeat transfers that have failed. If the problem persists, it can help to increase the target plate pinning pressure. With the Singer RoToR, it helps to avoid pinning errors if plates are consistently placed and pushed into the same corner in their holder.

Note 7: Quality control: image analysis

Pyphe-analyse produces a qc image for every image analysed. It is crucial to look at the images, even for huge datasets, to make sure colonies have been correctly identified and correctly matched to their grid positions. Issues with colony detection can usually be remedied by adjusting the threshold parameter (using the --t parameter), setting a hard threshold (using the --hardImageThreshold parameter) or using local thresholding (activated with the --localThresh parameter). The last is especially recommended for images with uneven brightness, e.g. those obtained with the Singer Phenobox. By default, pyphe-analyse excludes very small objects. If your colonies are very small, please adjust this exclusion parameter using the --s or --hardSizeThreshold parameters. For gridding issues, it can be worth to switch from automatic grid placement to manual one. Please see the pyphe-analyse documentation.

Note 8: Normalisation strategies

Pyphe-analyse gives the user several options for normalisation strategies. If neither the --rcmedian nor --gridnorm options are set, no normalisation is performed and the raw data from the plates is simply aggregated and summarised. For screens without a reference grid, pyphe can still be used (with row/column normalisation only or no normalisation). However, lowest noise values are obtained if both options are set. In that case, grid normalisation is performed followed by row/median column normalisation. This second normalisation can correct an artefact of the grid normalisation method which slightly over-corrects colonies next to the edge. This is essentially because colonies just off the edge are compared to colonies on the edge and therefore appear relatively smaller (see Figure 1—figure supplement 2B and Appendix 2 of (7)).

However, row/column median normalisation can and should only be used if the majority of strains in each row and column show no effect (i.e. the null effect can be reliably estimated by taking the median). This is usually the case for library screens where most of the mutants behave like wild-type and there are only a few outlying ‘hits’. For wild strain libraries, this case is harder to argue but it could still work if your strains are arranged randomly. It certainly will not work if your grid strain grows differently to the rest of your library (because only some rows/columns have a lot of replicates of the grid strain). In those cases, performing row/column normalisation after grid normalisation will do more damage than good. Pyphe-analyse produces qc plots for every plate analysed and you should inspect these carefully to check that the normalisation is working as expected.

Note 9: Quality control: biological and technical noise

Biological noise, technical noise and experimental errors (incorrect plate preparation, mis-labelling, pinning errors) will impact your data and can result in wrong conclusions if they go undetected. We therefore highly recommend to perform extensive quality control to quantify the unexplained variation and spot experimental errors. The use of negative-control positions (footprints) is key and plates with contaminated footprints should be discarded from the analysis. Furthermore, we include a number of control strains on every assay plate. One easy way to achieve this is to include an additional 96 grid of control strains on each assay plate. These colonies are not used during reference grid correction but are expected to have fitness values of 1. Based on these internal controls, it is possible to calculate two key noise indicators: First, the coefficient of variation (CV), the ratio of the standard deviation of the corrected fitness of these control colonies to their mean is an excellent indicator of the level of noise present in the assay. We usually exclude all plates which exceed a certain CV threshold (e.g. 10%). Secondly, the fraction of unexplained variance (FUV) is the ratio of the variance of the control strains and the variance of all strains. I.e. an FUV of 1 indicates that the spread of values is equally broad in the control as it is across the library, which can indicate that the observed variation across the library strains is purely technical. A suitable cut-off for exclusion of individual plates and conditions will depend strongly on your library. Certainly, an FUV of greater than 1 would be highly unusual and deserved further investigation. It can also be of value to exclude plate and conditions which show very small uncorrected colony sizes. This would indicate that the stressor included was too strong or the nutrients did not support any growth. In these cases, grid correction can introduce artefacts as small colony sizes are extremely noisy. These QC steps differ greatly between experiments, so they need to be performed manually on the pyphe-analyse long data report, removing spurious lines or setting them to NA. Once completed, you should proceed with hit calling using pyphe-interpret.

Acknowledgements

We would like to thank Natalie Barthel for critical reading of the manuscript. The study was supported by the Francis Crick Institute, which receives its core funding from Cancer Research UK (FC001134), the UK Medical Research Council (FC001134), and the Wellcome Trust (FC001134), and received specific funding from the Wellcome Trust (200829/Z/16/Z) and the ERANET COBIOTEC network “SyCoLim” under grant agreement BMBF-031B0931.

Conflict of Interest Statement

The authors declare no conflicts of interest.

Download PDF

Version 1

posted

You are reading this latest preprint version

High-throughput, high-precision colony phenotyping with pyphe

Status:

Version 1

Abstract

Figures

Introduction

Materials

Experimental

Scanning and Data Processing

Methods

General Method: Endpoints

Experimental

Analysis

Variation 1: Time course analysis

Variation 2: Colony viability analysis

Conclusion

Box 1: Other Software Resources

Notes

Note 1: Experimental design: power calculations

Note 2: Experimental design: plate layouts

Note 3: Preparation of layout files

Note 4: Preparation of suitable agar plates

Note 5: Preparation of the EDT table

Note 6: Quality control: pinning

Note 7: Quality control: image analysis

Note 8: Normalisation strategies

Note 9: Quality control: biological and technical noise

Declarations

Acknowledgements

Conflict of Interest Statement

References

Status:

Version 1