CU Boulder researchers across all STEM disciplines offer summer projects for SMART students.  In your Statement of Purpose, be sure to describe research areas/topics you’re interested in; this will help us identify research labs on our campus that align with your interests.  Here are some examples of the wide variety of projects available in summer 2023.

Genome Wide Association Studies of alcohol behaviors found a “risk allele” associated with higher levels of human alcohol consumption (a single nucleotide polymorphism (SNP) in the gene for the glucokinase regulatory protein). Our lab developed a mouse with this risk allele using CRISPR. We showed that female mice carrying this allele consume more alcohol than controls.  Summer SMART students will do behavioral testing of these mice (such as performance on a balance beam) to study their sensitivity and tolerance to alcohol. Experiments will involve injecting mice with alcohol, having them perform the behavioral tasks, and recording the data. Blood is taken and is analyzed for blood ethanol content, and the mice will be genotyped using polymerase chain reaction (PCR). Students will learn how to handle mice, perform behavioral testing and blood ethanol assays, and determine genotypes; and how to collect, record, analyze, and interpret the data.

Earthquakes can cause soil liquefaction, causing soils to lose their strength and stiffness. When this occurs in soils underneath buildings, it can cause extremely damaging effects on the built environment. For example, during the 2010-2011 Canterbury Earthquake in Christchurch, New Zealand, nearly 50% of the central business district was damaged by soil liquefaction, affecting 60,000 residential buildings. Seismic soil-structure interaction and stratigraphic variability and layering can influence soil liquefaction and the resulting damage to infrastructure. This research aims to better understand the physics of earthquake-caused soil alterations through physical experiments and numerical simulations.  

The student will perform centrifuge experiments using a new system for particle image velocimetry (PIV) analysis designed at CU Boulder’s 400g-ton, 5.5m radius centrifuge facility. They will analyze the centrifuge data and image processing of cameras’ recordings, and process data from the numerical simulations. The student will gain an understanding of centrifuge modeling, sensor calibration, finite element analysis, and Matlab programming.

Our research focuses on understanding the development and predictability of atmospheric environments conducive to the occurrence of extreme weather events. Current research topics include (1) diagnosing situations in which the traditionally separate polar and subtropical jet streams in the upper troposphere merge into a single, intense jet stream, (2) forecasting the type of precipitation during winter weather events (i.e., rain, snow, sleet, and freezing), particularly in areas of complex topography such as the Colorado Front Range, (3) understanding the development of extreme precipitation events in polar regions and their impacts on the Antarctic ice sheet, and (4) investigating the predictability of the upper-tropospheric jet stream and its influence on North American weather conditions. Students with a strong interest in understanding extreme weather events would be particularly well suited for this research.

Membrane technologies are important to address societal challenges including water scarcity, climate change and public health.  For example, reverse osmosis membrane is the leading technology for sea water desalination, while ultrafiltration and microfiltration membranes are important for virus clearance in biomanufacturing.  Our lab focuses on improving the performance of these membranes by using innovative methods such as surface patterning and stimuli-responsiveness.  For example, we develop acoustically responsive microstructures that can improve the antifouling performances of the membranes on demand.  The SMART student will conduct research on fabricating these novel structures, characterizing their responsiveness under acoustic excitation, and quantifying their impacts on the performance of the membranes under crossflow conditions.

Man-made climate change has influenced and will continue to alter the ocean.  Both the physical state of the ocean (e.g., its temperature) and the biogeochemical state of the ocean (e.g., its acidity, productivity, or oxygen content) have climate change expressions.  Sometimes, it can be hard to find these expressions above the background, natural, year-to-year variations in these quantities.  This project aims to use observations and climate models to examine year-to-year variations in ocean biogeochemical state, and to quantify the time when climate change expressions in these quantities can be reliably detected.


The ocean plays a crucial role in the global carbon cycle by sequestering atmospheric CO2. Using computational approaches, we study the interactions between carbonate chemistry and upper ocean physical dynamics on a global scale.  We can gain an idea of the strength of interactions between chemical and physical dynamics by looking at the Damköhler number (Da), a nondimensional parameter describing the ratio between the characteristic chemical and advective time scales. In this project, a SMART student will use a clustering method such as K-means (in MATLAB or Python) to study the structure of the global Da field. The student will first look at the time scales separately, then determine how other components that affect the time scales (e.g. wind speed, sea surface temperature, heat flux) are involved. This work could give us more insight into which regions and variables lead to strong interactions between the physical dynamics and carbonate chemistry in the oceanic mixed layer.

Down syndrome is the most common human developmental disorder and is caused by an extra copy of chromosome 21 (Trisomy 21). About 75% of Down syndrome embryos die by natural causes while in utero. We hypothesize this selective pressure may be related to differences in expression levels of the genes on chromosome 21. High expression levels of certain genes on chromosome 21, resulting from the additional gene copy, may be lethal to embryos. If this is true, then surviving individuals with Down Syndrome are expected to have more low-expressing variants of these genes than seen in the general population.  This project will examine data from 400 Trisomy 21 genomes and 700 control genomes and computationally determine if the Down Syndrome population contains statistically more of the low-expression alleles.

p53 is the most commonly mutated gene in cancer. It encodes a transcription factor that it binds to DNA and activates the transcription of target genes. The cell’s initial reaction to p53 activation is to transcribe small RNAs, called eRNAs, proximal to some p53 binding sites. Those p53 binding sites that produce eRNAs activate target genes that function to protect the body from cancer.

One project is to link each p53 target gene to a p53 motif/binding site that activates it. These links may be identified by correlating p53 eRNA transcription and target gene expression, 2D distance or 3D linkage. Using data science, the project will study the location of each p53 binding site in multiple cell types, determining if each p53 binding site has an eRNA and deciding if they are linked to a target gene. The goal is to build a prediction model that could guess which genes respond to p53 in any cell type. 

A related project looks at the evolution of this crucial pathway across species. It asks if the target genes evolve faster than the p53 eRNAs and if p53 eRNAs can move locations while activating the same target genes. We have found several target genes that are specific to certain sets of species. We hypothesize that they have gained a p53 binding site with an associated eRNA. With these insights we can explore these novel target genes and their regulation.


One of the primary uses of sequencing assays is to precisely identify the genomic locations of biologically relevant “regions of interest”. Precision is essential when, for example, researchers try to identify transcription factor binding sites from ChIP-seq, open chromatin regions from ATAC-seq, or the locations of unannotated RNA from various RNA sequencing assays. However, there is inherent uncertainty in any single sequencing experiment. To increase accuracy, researchers should consider the information from multiple biological or technical replicates to precisely identify these regions of interest. What is the best way to combine the signals from individual experiments to arrive at a “consensus” location? We have developed muMerge, an algorithm that uses a statistical framework to identify the most probable regions of interest from a set of experiments to make an informed decision about the consensus region. Currently, muMerge functions one region at a time on a single CPU. Significant software engineering is needed to optimize its performance for high-performance computing environments and add new software features that broaden its utility. We are looking for a student that would like to address these software engineering challenges making muMerge stronger, faster, and more versatile.

Genome Wide Association Studies (GWAS) of tobacco smokers have identified several genes associated with smoking behaviors. To better understand how these genes work, we use a cell culture model for behavioral testing. We transfect cultured mouse astrocytes with CRISPR DNA plasmids to study the effects of several genes, then assess the cells for dendritic branching using fluorescence microscopy. Once we find genes that affect the astrocytic response to nicotine, we plan to develop a mouse model. For example, astrocytes from mice lacking the Akt2 gene and exposed to chronic nicotine showed morphology consistent with intermediate astrogliosis. Astrocytes from wild type (WT) mice showed the opposite trend. With the mouse model we will perform behavioral testing including conditioned place preference, conditioned taste aversion, and prepulse inhibition.  SMART interns will assist with handling mice; perform behavioral testing, fluorescence staining, cell culture, and genotyping; and learn proper data collection and analysis.

Our laboratory uses a wide range of methods to study the molecular genetics, epigenetics and behavioral genetics of nicotine addiction and psychiatric conditions that often co-occur with nicotine use. The lab uses molecular methods, including CRISPR, to study how specific changes in the DNA of specific genes of interest affect gene function and expression in cultured cells. In addition, mouse models in which specific genes have been deleted or altered are studied to assess the role of the gene in various behavioral responses to nicotine and nicotine withdrawal. Also, we study the consequences of nicotine consumption by pregnant mice on the risk of behavioral disorders and alterations in brain biochemistry in the offspring and grand offspring of the nicotine consuming mother.

A formal group law is a power series F(x,y) in two variables x and y that satisfies certain properties akin to the properties of an abelian group. For example, F(x,F(y,z)) = F(F(x,y),z) for variables x,y,z, corresponding to associativity, while F(x,y)=F(y,x) corresponding to commutativity. Two examples are F(x,y) = x+y and F(x,y) = x+y+xy. Most examples are not this simple and in general F(x,y) is a true power series in the sense that it does not have a finite expansion in monomials x^ny^m. Just like groups, we can define homomorphisms between two formal group laws F(x,y) and G(x,y): A homomorphism is a power series f(X) such that f(F(x,y)) = G(f(x), f(y)). The automorphisms of F(x,y) are the invertible homomorphisms from F(x,y) to itself. The collection of all automorphisms from F(x,y) to itself forms a group, denoted Aut(F(x,y)).  Indeed, if f(X) and g(X) are automorphisms, we can compose them to form a new automorphism f(g(X)) and can check that this composition gives a group structure on Aut(F(x,y)). In this project, we will explore questions about the structure of the groups Aut(F(x,y)) for certain special choices of formal group laws F(x,y).

Pre-requisite: An undergrad course in abstract algebra

The prefrontal cortex is an evolutionarily advanced brain region that is critical for a wide array of higher-level cognitive processes that includes proper emotional learning. Dysregulation of prefrontal cortex function is associated with emotion-related psychological disorders, such as depression and anxiety disorders, including post-traumatic stress disorder (PTSD). Our research has found that neurons within the prefrontal cortex of rats have operational molecular clocks that express 24 h rhythmic activity (i.e. circadian rhythms). These “local clocks” are the product of intrinsic neuronal clock gene oscillatory expression patterns. We have found that rats have circadian rhythms in their ability to learn an emotional task. Manipulations that disrupt the function of the prefrontal local clock prevent this circadian rhythm of emotional learning. This type of emotional learning is impaired in individuals with PTSD. In order for us to better understand the underlying mechanisms by which local clocks in the prefrontal cortex regulate emotional learning, we need to directly manipulate clock gene expression within specific neuronal circuits in the prefrontal cortex of adult rats. We are developing an innovative viral vector/CRISPR-Cas9 strategy to knock-down expression of a local molecular clock present within specific neuronal circuits in the prefrontal cortex. The SMART student project will involve all aspects of using this strategy to study the role of prefrontal cortex local clocks in emotional learning that includes training in use and deployment of viral vectors in rat brain, behavioral assessment, gene expression monitoring, microscopy, and image analysis.