Inferred from High Throughput Mutant Phenotype (HMP)
HMP: Inferred from High Throughput Mutant Phenotype
- The HMP evidence code covers those cases when the function, process or cellular localization of a gene product is inferred based on differences in the function, process, or cellular localization between two different alleles of the corresponding gene in a high throughput experiment.
- The HMP evidence code is equivalent to the IMP code and the general guidelines for annotating with IMP code should be adhered to.
- Notes on what qualifies as high throughout data and general annotation guidance for high throughout experiments can be found on the HTP evidence code page.
Use of HMP
In high throughput perturbation screens, often particular phenotypic traits will be used as readout. It is particularly important that the phenotypic output assayed directly relates to the conclusion/annotation. Curators should be particularly careful about the choice of term and satisfied that the authors have taken sufficient steps to demonstrate direct correlation. A potential source of false positives may be from the mis-association of a phenotype to gene. For RNAi studies this arises from off-target effects and with mutagenesis studies whether the genetic lesion has been mapped to the correct gene. The curator should check that the authors have taken sufficient steps to reduce misattribution of phenotypes.
Features of a mutagenesis screen that may be suitable for annotation:
- Good correlation between phenotype and process.
- High confidence that genes targeted are correctly identified.
- Clear phenotype and high threshold for positive scoring.
- Reproducible phenotype.
- Verification by independent screening methods.
- Identification and exclusion of common contaminants/housekeeping genes.
Note: High numbers of potential annotations should not be used as a absolute indicator of whether an experiment warrants a HMP code. Sometimes large-scale genetic screens are aimed at comprehensively dissecting the process rather than a hypothesis-free probe. A detailed case-by-case review needed by curator to determine the methods and intent of the experiments.
Guidelines for high throughput RNAi experiments
In many high throughput instances, RNAi is used to disrupt gene function. RNAi screening experiments may be conducted on a whole-genome basis or on a targeted basis, where specific criteria are used to selected candidate genes. A major problem with annotating genes from an RNAi screen is the potential large proportion of false positives from off-target effects. Off-target effects result from the down-regulation of non-target genes or stimulation of other cellular pathways (e.g. antiviral response in mammalian cells). It is particularly important when evaluating a dataset generated by RNAi that sufficient steps have been taken to reduce off-target effects.
- Never annotate the primary screen from a whole genome or targeted RNAi screen.
Below are features of an RNAi screen that may be used to reduce false positives. Many are not absolute criteria for exclusion or inclusion, but should help a curator to judge whether the set is suitable for annotation.
- Takes steps to limit off-target effects e.g. uses multiple siRNA species for each target, rescreening of hits with different siRNAs
Includes positive and negative controls.
- Establishes a high threshold for positive scoring. If quantative, at least 2-3x greater than the mean (although many will use a negative control as the baseline). For high-content screening, using multiple parameters to define positive hits is generally regarded as increasing hit quality.
- Tests the signal-to-noise robustness of the assay. For some RNAi screens a Z-factor (or Z′) may be calculated from the standard deviation and signal response of multiple replicates of positive and negative assays (PMID:10838414). This gives a measure of the variability and dynamic range of the experimental setup. For annotation, the minimum Z-factor value should be 0.5. (Note:this is not the same as a z-score, which is a measure of deviation from the mean for an individual data point).
- Must be performed at least in duplicate.
- Validates hits by an alternative strategy.
- Measures a phenotypic trait that is unlikely to result from off-target effects e.g. translocation of a reporter protein into the nucleus.
- Screen out siRNAs that generate other phenotypes which may alter the output of the assay e.g. cytotoxicity, proliferation defects.
References for good practice in RNAi screens: PMID:19621037, PMID:22553869, PMID:20349460, http://www.mmi.med.ualberta.ca/facilities/ScreenGuidelines.pdf. Other statistical measures are described in https://www.ncbi.nlm.nih.gov/pubmed/19644458 PMID:h19644458]
RNAi studies that have taken significant steps to reduce false positives should be assigned a HMP evidence code within the guidelines of its use.
Evidence and Conclusion Ontology
Last reviewed: February 23, 2018