Risk assessment
The initial risk assessment of an experiment should be based on what is expected to occur and what could reasonably be expected during the experiment. If a result occurs that is not expected then it will be necessary to review the risk assessment.
Risk assessments must be written from the point of view that they must be understandable by the reader, for example, an HSE inspector or individual laboratory workers.
Generic risk assessments (such as we have for some plant species) are acceptable, but any inserts used must be defined and briefly described.
In completing application forms and risk assessments it is important that when a statement is made the justification and rationale is also given. For example, a statement that a vector is disabled is not sufficient on its own. The statement should say why and how this has been done.
An environmental risk assessment is necessary in the case of all risk assessments for genetically modified organisms. It is necessary to emphasise the containment measures that will prevent the risk being realised rather than just stating "no problem exists if X escapes".
Examples of risk assessments are given in the HSE Guidance document on pages 100-106. These are also available on the web (from Sheffield University) as follows;

E. COLI  K-12 DERIVATIVE EXPRESSING HUMAN GROWTH HORMONE.

CONSTRUCTION OF AN ADENOVIRAL VECTOR WITH A MODIFIED TISSUE TROPISM.

GENETIC MANIPULATION OF FOOT-AND-MOUTH VIRUS (only available in the HSE Guidance document) 

GENETIC MODIFICATION OF PATHOGENIC PLANT RNA VIRUSES.

Control Measures, Monitoring, and Inactivation 

Containment
The new regulations on genetic modification emphasise providing barriers between the genetically modified organisms and the environment. These barriers can be physical (e.g. engineered containment or heat sterilisation), chemical (e.g. disinfectants) or biological (e.g. the use of disabled strains).
When considering the containment measures necessary for any class of work, the frequently used statement in Schedule 8 to the Regulations "required where and to extent the risk assessment shows it is required" must not be regarded as indicating that the particular measure is optional. If the measure is not to be implemented the risk assessment must clearly indicate and justify why it is not required.

Waste
All waste must be inactivated before disposal. If disinfection is used, as opposed to autoclaving, the degree of kill must be known. It is suggested that ALL GM waste (certainly including plants but perhaps not whole animals) is autoclaved. The routine regime in Biology & Biochemistry and Pharmacy & Pharmacology is 126^oC for 30 minutes (specified by MAFF - precursor to DEFRA).
If disinfection is used as a means of inactivation it is acceptable to use manufacturers data on the degree of kill. However, it is essential that conditions of use are the same as specified by the manufacturers and that the contact times are consistent with that advice.
For projects designated as Class 2;
* The degree of kill expected in any waste inactivation should be stated
* The frequency of inspecting and monitoring safety cabinets and autoclaves should be specified.

Emergency plans must be prepared if the risk assessment indicates that as a result of any foreseeable accident or incident involving a significant and unintended release of any GMO;

• the health and safety of anyone OUTSIDE the premises may be seriously affected or 
• (for GMMs only) there is a serious risk to the environment.

Examples of areas which would need consideration are;

1. In the event of fire, would the fire brigade need to use different tactics to avoid water carrying GMOs into the environment?
2. Would emergency vehicles need cleaning in a particular way to avoid transporting GMOs off-site.
3. Would the police need to follow special procedures if the premises are burgled, vandalised or attacked?

Documentation
All people carrying out genetic modification work are required to read the risk assessment for that work and to sign the record sheet at the front of that assessment.

GMO characterisation
In order to help with GM risk assessments the following notes have been collated by Adrian Wolstenholme to provide basic information on the risks associated with commonly used hosts and vectors. 
These notes relate only to the host strains and vectors. Determining the potential hazards associated with any gene or expressed gene product is the responsibility of the project supervisor: these may substantially affect the level of risk implied here. Though every effort has been made to make these notes accurate, this is not guaranteed and the responsibility for ensuring the accuracy of risk assessments remains with project supervisors. 
Commercial suppliers can be a useful source of information on vectors. Vector information is available from New England Biolabs, Clontech, Promega, Invitrogen and Stratagene. Nycomed Amersham have a document entitled "Recombinant Protein Handbook: Amplification and Simple Protein Purification" (it costs £10 - August 2007).

Bacterial Strains
K-12 derivatives of E. coli, which include commonly used strains such as C600, DH1, DH5, HB101, INV1, JM109, TB1, TG1, XL-1Blue and the commercial derivatives of them (but NOT BL-21), are non-colonising and disabled. They are equivalent to hazard group 1, as they are not pathogenic to humans or animals. They often have auxotrophic requirements (this may vary from strain to strain) which are unlikely to be satisfied outside laboratory cultures. They have very limited survivability in the environment. There was relatively little data on the level of disablement of BL-21, and it may need to be considered as hazard group 2, though the rec^- derivatives can be assigned to hazard group 1. However, if you intend to use BL21, you may find it useful to quote from ACGM Newsletter 30 as follows:
"Stemming from a HSE-commissioned study, ACGM Newsletter 30 (Nov. 2001) updated guidance on use of BL21 and its derivatives as hosts. This states (in paragraph 3) that "this information implies that BL21 can be considered broadly equivalent to K12 strains and that in most cases work which uses this host can be considered as a class 1 activity.……, but the cloning of a bacterial pathogenicity determinant into BL21 will need careful consideration and may in some cases warrant classification as class 2". Since this project involves…. " (continue as appropriate)

Plasmids
The pUC series of plasmids (and those like them, which includes many commercial vectors; this includes pAT153, pBR, pBluescript, pBS, pGEM, pCAT, pEX, pSP, pT7, pMAM, pMSG, pXT, pEMBL) are non-mobilisable. The selectable marker gene (Amp^R) codes for resistance to an antibiotic not in clinical use. (This may not be true for other antibiotic resistance genes such as Tet^R).
Phage lambda vectors such as l gt10, gt11 and derivative such as l ZAP are non-mobilisable, as are all M13 vectors in a host with a tra^-F plasmid (most, but not XL1-Blue, DH20, DH21, NM522, SURE).

Cell cultures
All eukaryotic cell lines can be considered as especially disabled hosts and thus hazard group 1, unless the cells can colonise the worker (for example, if they are their own cells) or they contain adventitious agents which are potentially harmful. So all non-primate cell lines are group 1. The vector used must not be able to infect other cell types. Oncogenes that immortalise primary cell lines are of higher risk, but once the oncogene has become integrated the cells can normally be considered to be Group 1. (See part 2 of the Compendium for detailed guidance).

Viral Vectors
Use a disabled vector if at all possible. Replication defective E1 Ad5 vectors are unlikely to cause disease, but this will depend on the nature of the inserted gene and the final recombinant virus. The majority of retrovirus vectors based on MuLV, FeLV, ALV can be assigned to level 1, depending on the inserted genes, the type of packaging cell lines and the properties of the final virus. Those based on BLV and non-primate lentiviruses need special consideration. Anything using a primate lentivirus or HTLV is group 3. Although baculoviruses cannot infect humans and polyhedrin^- mutants have decreased survival in the environment they should not automatically be assigned to group 1.

Viral vectors (such as adenovirus) lack certain virus replication genes and are propagated in cell lines that complement the defect. Such viral stocks MAY be contaminated with replication-competent viruses, generated by rare spontaneous recombination events in the propagating cell lines, or may derive from insufficient purification. These vectors should be handled at the same containment level as the parent adenovirus from which they were derived.

Phage lambda vectors such as l gt10, gt11 and derivative such as l ZAP are non-mobilisable, as are all M13 vectors in a host with a tra-F plasmid (most, but not XL1-Blue, DH20, DH21, NM522, SURE).

Detailed advice on the use of virus vectors is given in part 2 of the compendium.

Plants
The following list represents assessments which have been carried out on plant species commonly used as sources of nucleic acids for GM work and as hosts for genetic transformations.

Tobacco (Nicotiana tabacum) - Tobacco plants, and particularly the leaves, store the highly toxic alkaloid nicotine. Small quantities are also found in seeds and presumably other tissues. Nicotine is the final product of a multi-enzyme pathway the genes for which are unlikely to be cloned and expressed together in any of the hosts. It is highly improbable that this compound or any related substance could ever be synthesised in the hosts. Reference source: A Colour Atlas of Poisonous Plants by Frohne-Pfander

Cassava (Manihot esculenta) - This species is a food crop plant that is cultivated in the humid tropics. All parts of the plants contain cyanogenic glycosides, which can release hydrogen cyanide upon damage to the plant. Cyanine can produce toxic effects in humans and other species if parts of the plant are eaten without appropriate preparation. The cyanogenic glycosides are the product of complex multi-enzyme pathways, and therefore their synthesis within any of the cloning systems described here is highly improbable.

Rice (Oryza sativa) - This is a cultivated crop species used for food production. It does not contain any toxic compounds so far as is known. The seeds are only ever consumed after extensive heat treatment (boiling). It is inconceivable that any deleterious product could arise from cDNA or genomic library production.

Thale cress(Arabidopsis thaliana) - This is a cruciferous plant and probably contains several glucosinolates which are deleterious if consumed under certain conditions, however, other cruciferous plants such as cabbage are eaten in large amounts without deleterious effects. These compounds are products of complex pathways and it is therefore unlikely that such products could arise as a result of cDNA cloning in bacterial hosts. Reference source: Effects of poisonous plants in livestock. Keller et al. 1978.

Arabidopsis lyrata as for Arabidopsis thaliana.

Cardaminopsis arenosa as for Arabidopsis thaliana.

Arabis holboellii as for Arabidopsis thaliana.

Maize (Zea mays) - This is a cultivated crop species. The seeds are eaten by humans and the whole plant is used as animal feed without any undesirable effect. It is unlikely that any deleterious product could arise from cDNA or genomic library production.

Personnel proposals

The crucial aspect of relevant experience is the all-important section of the proforma. That is why (with 'Markup' activetaed in the Word document) we have given examples of what is needed.