diff --git a/DGEAnalysis.pptx b/DGEAnalysis.pptx
new file mode 100644
index 0000000..a64f4e0
Binary files /dev/null and b/DGEAnalysis.pptx differ
diff --git a/Progress Presentation.pptx b/Progress Presentation.pptx
new file mode 100644
index 0000000..df320a1
Binary files /dev/null and b/Progress Presentation.pptx differ
diff --git a/idea1.md b/idea1.md
new file mode 100644
index 0000000..44b9350
--- /dev/null
+++ b/idea1.md
@@ -0,0 +1,22 @@
+#DATA 400: Idea 1
+####Ryan West
+
+## Part 1: Research Question
+
+One of the largest fields of biology that uses data analysis is that of bioinformatics. One large field of bioinformatics is that of differential gene expression analysis, which compares the gene expression of specified conditions to typical gene expressions. The NCBI keeps records of large amounts of trial data. I chose to answer the question: *What affect does IBD (Crohn’s disease) have on gene expression when compared to healthy individuals?*
+
+## Part 2: Dataset
+
+The dataset that I’ve chosen comes from an RNA-seq analysis of adipose tissue-derived stem cells from individuals with inactive and active Crohn’s, along with healthy individuals. The dataset is available in an annotated form from the NCBI’s GEO database, or in its raw FASTQ format from the NCBI’s SRA database. I have yet to decide which format I will use, as both have their benefits and issues. 
+
+The raw FASTQ file contains the individual transcriptomes associated with each person who was analyzed in the study. This would provide a great opportunity to analyze raw data including its clean-up, but the files total around 15 Gb. The annotated file would skip some preprocessing and would be much smaller.
+
+The dataset includes the transcriptomes of the patients analyzed, along with basic information about each patient, such as sex, age, and cohort. There will need to be more columns added to compare each sample to each other, with their associated p-values.
+
+## Part 3: Differential Gene Analysis
+
+The main purpose of differential gene analysis is to determine which genes are significantly up/down regulated when compared to a ‘normal’ transcriptome. The data will need to be processed and normalized, which is usually done in the program FastQC. If any issues are identified, the data is typically trimmed using Trimmomatic. 
+
+The next step would be to align the samples to the human reference genome, which can be done in R using HISAT2. From here, we can get feature counts to be used in analysis. This data can be placed directly into R for analysis with the DESeq (or edgeR) packages, which have been designed to read gene expression.
+
+I will create various heat maps and charts to show the statistical analysis of which genes are significantly differentially expressed. Another chart which is typically created is a volcano plot to visualize up/down regulation. Typically, this information would be important in understanding pathways in diseases for a whole host of reasons, potentially for new drugs to treat patients.
diff --git a/ideas2-3.md b/ideas2-3.md
new file mode 100644
index 0000000..e632245
--- /dev/null
+++ b/ideas2-3.md
@@ -0,0 +1,35 @@
+# Idea 2: 1000 Genomes Project
+
+## 1) Research Question
+
+Where is hypertension most common, and what genetic characteristics are associated with developing hypertension? (This question could be asked about nearly any genetic condition; I have chosen hypertension.)
+
+## 2) Data Source
+
+The 1000 Genomes Project was conducted between 2008 and 2015 with the goal of documenting genetic variation across diverse populations. The human genome contains approximately 3 billion DNA base pairs; this project aimed to document the genome of 1000 people. 1KGP keeps its information freely available online for research purposes. This contains filterable versions of genetic material and can be grouped by world region. We can get information on known genetic mutations from ClinVar to know where in the genome (out of 3 billion bases) we should be looking.
+
+## 3) Discussion
+
+We can filter and trim this data down to the genes that we need to find when known malignant genes are overexpressed in the population groups discussed above. There are a few R packages to help with this (like PLINK), and further statistical analysis can be done in R (like Chi-square) to better determine allelic frequency. There are also pathway databases which could be used to illustrate how the specific genes we’re looking at are involved in cellular processes and what happens when variations are found.
+
+We could also create a heatmap showing which genes are significantly overexpressed in different regions of the world. There likely are not any ethical implications for this, as they have already been handled by 1KGP. There is no personally identifying information associated with anything in the data to be analyzed. This analysis could be helpful to find interesting patterns amongst populations and can better our understanding of where and why some conditions are prevalent.
+
+# Idea 3: Sentiment Analysis
+
+## 1) Research Question
+
+How can Utz Quality Foods better prioritize call center agents’ time?
+
+## 2) Data Source
+
+Utz Quality Foods’ Zendesk instance. All the problems that consumers have ever had with Utz going back approximately 5 years are stored. They are accessible via API and are delivered in JSON format, making analysis on them quite easy. This data contains many different topics, and there are around a few hundred variables. Large chunks of these variables are usually listed as NA or are blank, but analysis would likely be quite easy on this data.
+
+## 3) Discussion
+
+Utz’s customer data is currently quite messy. A large portion of this project would be cleaning up the data and constructing better workflows, with the potential for more powerful analysis later (perhaps too far outside of the realm of a mini project).
+
+My project would have two goals: route tickets to agents with the most specialized skills and better prioritize tickets. Currently, Utz’s consumer data receives requests from customers and dumps it all in one place with the same priority. It would be extremely beneficial to perform some type of sentiment analysis on the text body (combined with customer-selected options) to automatically assign priority labels to these requests so that agents are working on the most urgent requests first (while also keeping time in mind).
+
+Additionally, analysis of this text (again combined with selected variables/tags) would allow us to better group requests and deliver them to more appropriate agents. This will greatly improve organization for Utz.
+
+One huge concern is that of ethics: some of the data is confidential or at the very least sensitive, so this model would likely only be able to be shared internally or with portions of it redacted.