tweaks

Author: dpark01

docs/recommendations.md

@@ -89,7 +89,7 @@ There are a multitude of options for bioinformatic computing infrastructure depl
 
 Note that the abstraction level is independent of where the physical location of the compute resources are.
 
-The choice of abstraction level influences and is influenced by many other factors. Throughput, i.e. the number samples that need to be run through the system; flexibility, which includes to what extent analyses can and should be modified by users; scalability, which includes to what extent additional users can be added, and the extent to which the solution allows for changes in data volume; external access requirements, which includes the frequency data needs to be sent to external servers and proportion of time required connected to external servers; and sustainability, which includes to what extent financial and competent human resources are and will be available to support the chosen solution.
+The choice of abstraction level influences and is influenced by many other factors. **Throughput**, i.e. the number samples that need to be run through the system; **flexibility**, which includes to what extent analyses can and should be modified by users; **scalability**, which includes to what extent additional users can be added, and the extent to which the solution allows for changes in data volume; **external access requirements**, which includes the frequency data needs to be sent to external servers and proportion of time required connected to external servers; and **sustainability**, which includes to what extent financial and competent human resources are and will be available to support the chosen solution and the level of maintenance and upkeep required to be performed by lab staff.
 
 ![service-diagram-20211116-dp](https://github.com/pha4ge/infrastructure-resources/assets/8513746/d0c66464-28ba-4467-8f20-b0a39cc65d5c)
 **Figure 1: Scope of Responsibility** - the scope of what different solutions attempt to solve for the users and admins. All infrastructure components (bottom of figure) need to be addressed by some combination of the end-user, system admins, or an external party. The degree to which responsibility can be deferred to a third party depends on the solution chosen. For example, on the IaaS level, a research group will also need to provide their own solution for orchestrating infrastructure (including software installation, selecting and installing a pipeline executor and installing pipelines), which would be solved via SaaS. Thus, selecting a lower level of abstraction introduces additional flexibility with the burden of additional complexity for the users.