Dysregulated immune cell infiltration in Idiopathic Pulmonary Fibrosis pathogenesis
Advisor: Dr. Allendoerfer
Internship Location: ilab Research Instiute
Onsite Mentor: Qian Wang, ILab research institute
This project's goal is to understand the relationship between permanent chronic scarring of the lungs (Idiopathic Pulmonary Fibrosis, IPF) and the immune system. In normal lungs, the scarring of lung tissue is a protective immune response; however, usually there is the regeneration of tissue. In IPF this regeneration is lacking and the scarring initiated from the immune system is uncontrolled. As the lung tissue gets scarred it makes it difficult for the necessary gas exchange to occur.
Project Blog
Week 1: Blog Post
Hi everyone! My blog will be about a lung disease known as Idiopathic Pulmonary Fibrosis. Today's post will just be a brief intro to why I chose this project while including terms that I will expand on in future posts.
Why this project?
During the COVID pandemic, which I am sure everyone is familiar with, I began to really understand the importance of the lungs. I became curious about how the lungs worked and began to wonder about the various diseases that negatively impact the lungs. Through an online bioinformatics program, I stumbled upon the disease Idiopathic Pulmonary Fibrosis (IPF).
What is IPF?
IPF is a lung disease characterized by permanent scarring of lung tissue. In normal lungs, the scarring of lung tissue is a healthy immune response however in IPF this scarring is uncontrollable and there is no regeneration of tissue. This makes it incredibly difficult for necessary gas exchange in the alveoli and interstitium (parts of the lungs) to occur.
My goal
After reading a bit I realized that there wasn’t much known, especially in terms of diagnosis and treatment as many of the times it is hard to identify if the disease is IPF or another lung disease. This project's goal is to understand the relationship between the permanent chronic scarring of the lungs in IPF and the immune system, especially by analyzing genetic overlaps between the two. I am specifically doing this to better understand the molecular aspect of this disease and see if prevalent genes that may assist in more treatments can be found. Currently, there are a few genes that can be linked to IPF however the role immunology has in this disease is a bit more unknown. There is also not much known about genetic susceptibility to IPF. I believe I will find genes that are triggered during the inflammatory process and scarring of tissue while also finding fewer expressed genes in terms of tissue regeneration. In summary, this project will take a bioinformatic approach by using Bulk RNA and Single-cell RNA sequencing analysis to uncover associations between immune loci and IPF.
Before I go I will quickly give a brief intro to Bulk and Single cell RNA sequencing. Bulk RNA sequencing is a type of analysis that is used to measure the average gene expression levels across tissues. Single cell RNA sequencing is a lot more specific compared to Bulk RNA as it analyzes individual cells within a tissue sample. For both analyzes I will be using R studio as a platform to run my code.
This was a lot of information all at once, however, I hope in the next post I can clarify more and dive more into the literature review! Hope to see you next time
Thanks
Pragyaa
Week 2: Blog Post
Today’s post will have a lot more jargon as I will try to explain IPF in a lot more detail while also explaining how I will contribute to this field. I hope you are ready to learn a lot more about the lungs and feel free to ask questions as this post will likely be slightly more complex
Introduction
IPF also known as Idiopathic Pulmonary Fibrosis is a lung disease that falls under the broader term of diseases known as ILD (Interstitium Lung disease). ILDs are a group of parenchymal lung diseases characterized by varying degrees of inflammation and fibrosis (inflammation is swelling and irritation while fibrosis is scarring of lung tissue). These diseases can be acute meaning there is a short-term or rapid response and chronic meaning a more long-lasting response. Most of these diseases are caused by infection, exposure to particles, and genetic predisposition with IPF being the most common disease within ILDs.
IPF is a chronic disease that is formed by the thickening of lung tissue forming scars in the lungs. In healthy lungs, oxygen passes easily through the alveoli however the scars in IPF make these walls thicker making it harder for oxygen to pass through. IPF tends to be incredibly difficult to diagnose especially in earlier stages due to the lack of noticeable symptoms until the disease progresses and the similarities in symptoms to other lung diseases. That being said, this disease can be characterized by the presence of a radiographic and/or histopathological pattern of UIP (scarring in a honeycomb shape). This usually starts at the base and edge of the lungs gradually progressing to involve all lung tissue.
Epidemiology & Specific Mechanisms
Normal lungs are normally exposed to lots of injuries but they can repair themselves however in IPF this mechanism is missing. When this repair mechanism doesn’t work this could be because of ongoing activation of mesenchymal cells (multipotent stem cells involved in tissue regeneration, homeostasis, and many more), excessive production of pro-fibrotic mediators (contribute to fibrosis needed for wound healing), premature + persistent epithelial cell senescence (mechanism that stops cells with unstable genomes from being cancerous but can contribute to aging diseases), or stochastic pro-fibrotic epigenetic reprogramming (altered gene expression that leads to fibrosis).
Cellular & genetic mechanisms
At a more cellular level, IPF is an epithelial-driven disease where unnatural activated lung epithelium produces mediators of fibroblast migration (leading to excessive fibroblasts present in lungs), proliferation, and differentiation into active myofibroblasts (leading to excessive fibrosis). All in all in IPF there are several cellular mechanisms leading to excessive scarring and fibrosis. One major cell involved in IPF is called alveolar epithelial type 2 cells (AEC2s). In IPF these cells tend to have genomic instability, telomere attrition, epigenetic changes, loss of proteostasis (cellular process to maintain structure and function of proteins), deregulated nutrient sensing (cells' ability to respond to nutrients), mitochondrial dysfunction, cellular senescence, and altered intercellular communication.
In terms of genetics, a few important genes are TERC, TERT, TINF2, DKC1, RTEL1, PARN, and NAF1 and each one plays some role in telomere maintenance. One other important gene is MUC5B. This is one of the most common polymorphisms that increases the risk of IPF. It is however unclear how MUC5B relates to IPF though. Several important anti-fibrotic genes are silenced leading to the loss of prostaglandin E2 (important anti-fibrotic mediator). There are a few other genes that have been discovered to be altered in IPF but it is unknown how some of these affect IPF.
Bit about Immunology
I know there has been a lot of technical talk and I promise I am almost done. The final section will cover immunology. Most of the research has surrounded T cells which protect the body from pathogens and cancer. In lungs, CD4+ and CD8+ T cells are the dominant adaptive immunocytes (white blood cells part of the adaptive immune system) and these cells are usually long lived memory T cells which provide protective immunity & type of memory cell that is in the tissue. It has been shown that CD8+ T cells accelerate lung damage and higher levels indicate more severe lung injury. The biggest conclusion however is imbalance between inflammation and inflammation resolution leads to fibrosis.
Existing Research
While it remains unclear, a study says that immune characteristics such as recurrent inflammation and immune system dysregulation may be a cause for IPF. According to the same study, 2 distinct immune phenotypes among IPF patients were inflammation-activated immune cell infiltration phenotype and immune-suppressive phenotype. Inflammation usually is the immune response to an injury/infection to fight off infection however in IPF usually this is uncontrolled and prevents the lungs from performing their usual functions. The suppressive phenotype is when the immune system is weakened.
Next, there has been research in immune cells, specifically T cells and immune stem cells. Mesenchymal stem cells (MSC) play an important role in modulating immunity and in lung regeneration. In fact there have been clinical trials to prove that MSC can be used as a treatment, and while some have been successful there needs to be more trials to prove the efficacy of MSC therapy in IPF patients. Activated immune cells stimulate fibroblasts to secrete extracellular matrix initiating lung tissue repair and regeneration.
In terms of T cell-related genes and proteins (e.g. CD28 and LCK) are all biomarkers in IPF. It was proven that a significant increase in CD8+ CD28 null T cells was present in IPF lung tissues. CD28null T cells may potentially promote fibrosis, however, the immune checkpoint CTLA4 may be protective and limit this effect of fibrosis. Despite this there is still some unknown about T cells specifically how T lymphocytes are associated with IPF.
My Contributions
My project however is more focused on not only the genetic variation between normal and IPF lungs but also the pathway interaction between various genes. Finally, my project will also explore the overlap between the immune system and this disease to determine which genes are involved in both. Hopefully, with my results, we can better understand the role genetics and the immune system plays in IPF, especially by exploring the various pathway interactions.
This is the end of my quite lengthy post so thank you for staying and reading. I can’t wait to start sharing my progress in my next few posts.
Pragyaa
Week 3: Blog Post
Hi everyone! Today's post and those that follow will begin to entail the methodology I am working on. Currently I am working on Bulk RNA sequencing analysis which is a technique used to compare gene expressions between healthy and diseased (in my case, IPF) tissue. This week, my goal was to complete Sequence reads, FASTQC, and Adapter trimming, which are all ways to help pre-process the data before it is ready to run more advanced code. Sequence reads are short sequences of RNA generated by RNA molecules that are analyzed to identify differentially expressed genes (genes that have significantly different expression levels between 2 groups). FASTQC is a file type that has the sequence reads and adapter trimming is a bioinformatics tool that removes DNA fragments added during library preparation.
After researching what each term meant and the functions that entailed each I began to try to figure out where I can find the FASTQC file. After doing hours of research I realized that my data was already pre-processed so I didn’t have to perform sequence reads because none of the data was in the FASTQC format. That being said, I ran into a new problem.
Manual Data Processsing
I tried downloading the raw data and normalized data but it downloaded as a zipped file. This means that the file was encrypted and it is a format that isn’t readable. This led to me researching the commands on how to unzip files in the command prompt and this took me quite some time as I didn’t know the functions. Once I unzipped the file I realized it was in txt format. Initially this wasn’t a problem however I realized that the data had information for IPF, ILD, CLAD, and normal however I needed to remove CLAD and ILD. The format of all the samples was also not in an identifiable form meaning that they were all in ALF0 with a number. I had to then go into each sample (there were 139) and change their names into IPF#, Normal#, CLAD#, and ILD#. Once the samples were all renamed (this took a really long time) I copied the names and changed the txt file. I wasn’t done yet though. Files in txt format are not cleanly formatted meaning they are not in rows and columns making it hard for me to remove the CLAD and ILD data. To make the data cleaner I converted it into a csv file by opening it in excel and removing the columns.
Coding
Finally my data was ready for code. I downloaded several packages I needed such as enirchplot, reshape2, and ggplot2. I made sure the code could properly read the data and created boxplots to check if the ggplot2 package works while also making visualizations of the data. Finally I imputed all the samples names (the ones that I changed into IPF# or Normal# as well as their group, IPF or Normal) and now my data is ready for mapping where I will be using another package to figure out what genes are in this data.
The data I am using for my Bulk RNA sequence analysis is from this link:https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE213001
Thank you so much for reading and I hope to see you in my next post which will likely include more visuals and results for Bulk RNA sequencing.
Week 4: Blog Post
Hi everyone! Hope you all are having a nice day so far. Today’s blog post will mostly cover gene mapping while also explaining my entire week of struggles with it.
Explanation
When the Bulk RNA dataset was downloaded it had rows of ensembl IDs. Ensembl IDs are unique ids assigned to, in this case, genes. That being said, for human data (which is what my project is focused on) they tend to come in the format of ENSG (eg. ENSG00000182378). The point of gene mapping is to match these Ensembl IDs to a gene name as these IDs don’t give as much information as the gene name could. In order to do this there needs to be an annotation file. Annotation files can have loads of genetic information however the ones that pertained to me were ensembl ID and genetic name. In order to try to match and merge the files I needed to download packages such as rtracklayer.
My Process
Now that I have explained the purpose and a bit of how to pull this off I will delve deeper into the several problems that I ran into. The biggest problem was finding an annotation file that had all my ensembl IDs matched to a gene. I tried using data from UCSC, GENECODE, USEast Ensemble (tried USWest, asia and many other mirrors), and genenames. For some websites there were different versions so I downloaded several versions to try and test if they had all my ensembl IDs matched. Before I began merging the 2 files I had to convert the rows in my data set (norm.exp) to columns as the ensembl IDs were in rows. If the ensembl IDs were in columns then it would be easier to match the gene IDs to the ensembl IDs. The general pattern was once I downloaded the file (some of them were incredibly large and took time while other I had to copy and paste to a separate document as downloading didn’t seem possible), then I had to unzip the file, check the data (and if necessary re-format them), sometimes change the names of the columns, merge them, check the merged data to see if there are actual gene names, then check if there were any ensembl IDs missing and if they were I would print them out to see which ones. Sometimes the format of the ensembl IDs from the annotation file was different compared to norm.exp as there may be decimals in the ensembl IDs. To take care of that I had to see a portion of the downloaded data and if there were in another format I would have to convert them. This is the simplified pipeline of the procedure but as someone new to coding it did take me a long time because of the numerous errors I ran into (the most frustrating one was the gene names all had NA).
Sucess! (Finally)
Today I was finally able to find an annotation file that matched all of the ensembl IDs meaning that now all these IDs are matched to a gene. Due to the time it took for me to figure out and find the files I wasn’t able to fully complete the Bulk RNA sequencing like I planned to, however the next steps just include making visual graphs and charts like heatmaps. Next week's post will definitely have the results of the Bulk RNA sequencing and likely also include the introduction of the more complicated analysis: single cell RNA sequencing.
Thank you all so much for reading today's post! Hope to see you next Friday.
Pragyaa
Week 5: Blog Post
Hi everyone! Today’s post will include a lot of issues I ran into (and maybe a bit of ranting) because this week sure did not go as planned AT ALL.
Review of last week
Last week I ended with merging the annotation file with my normalized dataset which I was going to use to perform the rest of the analysis. This week I worked on trying to generate visual graphs and get more statistical results. Before I get into the several problems I ran into I will give a summary of what I have done.
Summary of what I did
BCV Plot
First, I had to use DGEList, which is an object in edgeR (an R package) that helps with startical modeling to identify differentially expressed genes. This information is important because it helps with revealing the molecular process the gene is involved in, which includes how they are involved in certain diseases (in my case, IPF). Then, I removed the low-expression genes and used this data to create a BCV plot. A BCV plot, also known as Biological Coefficient of Variation, measures the variability of gene expression levels. The graph I generated shows IPF and Normal samples in pretty distinct clusters, meaning that their genetic expressions are different between the two. In the graph, points that are closer together have similar expression patterns, while samples farther apart are more different. If there is a clear separation (which I had), then that means that there are significantly differentially expressed genes.
PCA Plot
I also generated a PCA plot also known as Principal Component Analysis. This plot had the X axis labeled as PCA1 and Y axis labeled as PCA2. PCA1 represents the maximum variance, so the further along the axis, the more variance. PCA2 captures the 2nd most source of variance, and if there was PCA3, the trend would continue. The IPF samples are more spread out compared to normal samples, meaning there is more variation with IPF. The larger the number is (whether positive or negative), the higher expression value that drives the separation between the samples.
From there, I had to determine the up-regulated and down-regulated genes. To do this, I had the FDR and logFC values to help separate the samples.
Heatmap and gene ontology
The last few steps were to make a heatmap and run gene ontology. For the heatmap, unfortunately, I wasn’t able to figure out the most differentially expressed genes (where majority of my problems came from). I thought running the gene ontology might help narrow my search; however, before I could complete it, my code crashed my device (I’ll get into that later). Gene Ontology Enrichment analysis is used to identify biological processes, cellular components, and molecular functions that are statistically significant/overrepresented in a set of genes. Currently, I have only generated this analysis for the upregulated genes; however, I have bar graphs and cnetplots to highlight the important pathways within the upregulated genes in my dataset.
Problems!
Now it's time for me to complain about the problems! So, the first issue was that DGEList can only use raw data; more specifically, it can’t process negative data, which usually comes from normalized data. This meant that I had to relabel my 139 samples and sort them. Originally, I thought I could just copy and paste the samples I had from the normalized data; however, I realized that there was a mistake (2 of the samples had the same label), so I had to rename everything. The other plots were fine until I reached the heatmap. I spent hours trying to figure out if there was a way I could find the important genes. I tried to create a map for the top 100 then 200 to see if increasing the gene samples I tested would help (I ran into some issues with this so I am still not sure if it works or not, I have to test it again), I tried to see if I could sort by highest variance (didn’t work), and I tried to sort with logFC which also failed. The final one I can try is to manually put in the genes to map, but to do that, I need to figure out which genes are the interesting or important ones, which is something I am not sure how to do. I am hoping that from the gene ontology, I can maybe try to eliminate a few, but I doubt it will help as it doesn’t seem to tell me about the genes compared to specific processes. Maybe I can try to find what genes are involved in that process, but I am not sure. My final problem, which is the biggest, is that the code is getting to be too much for my device to process. Originally, I thought it was my laptop's problem because it was an old laptop; however, when I transferred the code to a different device, it still struggled to run past 1 line. I am not entirely sure why because it has been fine for the past two weeks. The line that it can’t seem to do is merge the annotation file to my data for the raw data (it can do the normalized one fine, though).
Anyway, I hope to solve these problems and try to wrap up the Bulk RNA analysis.
Thank you so much for reading!
Pragyaa
Week 6: Blog Post
Hi everyone! Today’s post will finally have the results from the bulk RNA analysis while also providing a brief introduction to the single-cell analysis.
Introduction
After looking through my code I realized that the rows of my dataset were removed and replaced with integers which was causing several problems later. As a novice in coding, I didn’t realize this problem because I didn’t preview the dataset until later so it took a while to figure out which line caused the error. Other than that the process was pretty smooth. I generated heatmaps and ran gene ontology analysis to better understand the pathways and genes involved in the samples.
Results
Figure 1: Heatmap
Figure 1: This is the heatmap that I generated with some of the samples (if I used all of them it would be incredibly crowded). The “x axis” has the samples while the “y axis” has the genes. The color and intensity of the boxes are used to represent changes (not absolute values) in gene expression. The color and intensity of the boxes are used to represent changes of gene expression across certain samples. As it can be seen blue is mostly for the normal samples and yellow (which is the other end of the spectrum meaning incredibly different gene expression) is for IPF samples. The numbers on the scale indicate the magnitude of gene expression levels across the samples.
The next step was to run gene ontology which helps with understanding the molecular functions and cellular processes about the disease I am specifically researching, providing me with a better understanding.
Figure 2: Cnetplot
Figure 2 shows the genes involved in certain pathways. The pathways are represented with the colored lines while the size of the nodes represents the number of pathways the genes are associated with. The genes that have more connections may play a central role in disease-related pathways that are listed on the graph. This plot is constructed with the enrichGO() function from the package clusterProfiler which checks which biological pathways are statistically enriched in the input genes. Through the heatmap it can be determined that there are some differentially expressed genes as there are striking differences in the colors and scales thus it is very likely that some of these genes and pathways are relevant to the gene. Since there is a control group in the data it is very likely that some of these genes highlight biological processes altered in this disease.
Figure 3: Barplot
Figure 3: This is the last important plot I will show and this just shows the pathways that are more statistically significant. The lower the p. Adjust value is the more statistically significant the pathway is in terms of IPF.
I generated a lot more graphs however these are 3 of the most important ones in Bulk RNA. Before I end today's post I will explain my single cell analysis briefly. Just a quick reminder Bulk RNA analysis looks at the gene expression across the entire cell population while single cell analysis focuses on individual cells.
Intro to single Cell
To perform single-cell analysis I needed to find another dataset (the dataset I used for bulk RNA collects the samples differently providing me with different data). When looking at the single-cell dataset (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE141939) and selecting a specific sample the way of collecting the data was different. The researchers collected and harvested epithelial cells using bronchoscopy. Then they prepared single-cell libraries and sequenced the samples on a sequencer. For this week in terms of single-cell analysis, I loaded the dataset and did part of the analysis by constructing tSNE plots and Umap plots that I will show and explain next time. Currently, these plots only serve as a reference however once I start with clustering and identifying the important cellular groups these plots will be more concise and meaningful.
Thank you so much for sticking around and reading till the end! Can’t wait to share my progress with you all next week (this part will likely be one of the most important steps)
Pragyaa
Week 7: Blog Post
Hi everyone, hope you all are having a nice Friday evening! Today’s post will be relatively short because most of the work I have done is more labor-focused compared to more conceptual.
This week, my goal was to cluster my samples and name each cluster. Initially, I wasn’t sure how to name each cluster, so it took some research to figure that out. I used the function FindAllMarkers() to group my genes into clusters and make a CSV file. From there, I had to make violin plots for each gene. The violin plot can tell me which genes I need to look at to determine what cell type the cluster corresponds to. A violin plot has the x-axis labeled with each possible cluster (for me, BR001…BR015) and the y-axis with expression levels. The purpose is to show in which cluster a gene is expressed. If the gene is significant, it would only be expressed in 1 cluster, and instead of a straight vertical line (which is possible for many genes), it will have a shape similar to a raindrop. This means that, unfortunately, I have to look at every single gene, more than 2000, and generate plots for each of them. So far, I have done about 800, however, I am realizing that according to the function, some of the genes are supposed to be a part of cluster 1, however, the violin plot shows them to be a part of 14. This might be because the clusters are similar and could be combined; however, I will have to make a heat map to determine if this is true.
Currently, I am still working on the mundane process of generating graphs, saving those graphs, and then later determining the important genes. From there, I will research what those genes do, and that should help me determine the cluster type. Hopefully, I will be done with these genes and will share pictures of my graphs
Thank you so much for reading!
Week 8: Blog Post
Hi everyone!
Last time, I left of by mentioning how I was going to try and identify/label the clusters. As I kept going the original way (looking at each gene and making violin plots to determine which cluster they are in) I realized that this was going to take months to do because I have 5000+ genes. After doing more research, I found a way to generate the top 10 differentially expressed genes in each cluster, meaning I had 170 genes to identify (I have 17 clusters). I used protein atlas and gene cards to search for the gene and determine in which cells they were mostly expressed. From there, I would search up the cells and determine what the cluster should be named. This led me to make this map.
This map clusters all the samples and labels each cluster. This tells me about the whole data in general however, it doesn’t tell me much about the disease I am trying to study. From here, I wasn’t really sure how to separate these clusters into IPF and normal to compare the genes and cells. After some research (I am still working on this part), I clustered the samples into IPF and normal clusters and sorted the cell types as well (this graph might not be completely right)
This clusters the samples into IPF and normal
This is the plot that I still need to work on, but the idea is to cluster the data into IPF and Normal, and hopefully, I can label these clusters by identifying the top differentially expressed genes.
This is the final plot I created, and it basically shows the distribution of IPF and Normal within each cluster
Currently, I am trying to label the clusters of the map I mentioned before, and I am trying to determine the top differentially expressed genes in IPF and Normal while also determining the top differentially expressed genes within each cluster.
I am slowly coming close to the end of this project and thus will hopefully have final results to share soon! Thank you so much for reading today's post!
Week 9: Blog Post
Hi everyone! Today's post will be relatively short, covering how I ended up with an Excel sheet sorted by clusters with the differentially expressed genes (DEGs).
To do this, I started with an if statement to sort the genes by logFC threshold. LogFC represents the differentially expressed genes, making it useful in determining the differences between the control and condition. This allowed me to generate a CSV file for each cluster with a list of genes, pval, avg_log2FC, pct1, pct2, and p_val_adj. I had CSV files for 17 clusters. Pct1 determines the percentage of expression in the current cluster, while Pct2 determines the percentage in the rest of the clusters. The larger the difference, the more important the gene is to the cluster it is assigned to. P val represents the probability of observing certain values if the null hypothesis is true, whereas p_val_adj is performed when many statistical tests are run, meaning they have an increased probability of finding false positives
I also made a generalized CSV file for my filtered DEGs, which sorts my genes based on the same characteristics I listed above. My p_val_adj <0.05, avg_log2FC > 0.5, pct.1 > 0.2 & pct.2 > 0.2. These were all the standards that I had chosen. Currently, I am trying to add a column for clusters in this CSV file, as this file seems to separate the genes by looking at much more, which is more advantageous as it’s taking more into account.
I have still been unable to name the clusters in the umap that I mentioned previously (the one that sorted the clusters based on IPF and Normal); however, I have been able to somewhat sort the genes based on the clusters to determine the differentially expressed genes.
Thank you all so much for reading!
Pragyaa
Week 10: Blog Post
Hi everyone! Today will be another short post as I am slowly reaching the end of my project and preparing to collect and analyze my results.
This week was mostly focused on labeling the graph that compared the IPF clusters with the Normal sample clusters to potentially help draw comparisons between the cell types involved. While this was simple (and I over complicated the process) I did run into a strange problem which I haven’t been able to understand the cause of. Currently, I have 16 clusters however a few days ago I ran my code and I got 17 clusters despite not changing a line. Then R ended up restarting and I had to re-run all my code from the beginning (which takes hours as there is a lot of data processing and clustering which takes time) and I somehow got 16 clusters again. I looked over my code several times to see if there was something I changed or something that looked right however I couldn’t find anything.
Another step I worked on was generating upregulated and downregulated genes for each cluster. This was pretty simple as all I had to do was filter through the genes that had a positive logFc value (upregulated) or a negative logFc value (downregulated) and made separate csv files for each. After looking through I have several 1000 for some so I need to see if I can try to filter them further or find a way to prioritize a few of them.
As I am reaching the end of my project my next step is to put all the results I got (including the Bulk RNA) and analyze each graph to form conclusions about my research topic and question.
Thank you so much for reading!
Week 11: Blog Post
Hi everyone, my next few blog posts are going to be focused on the results and paper that I am currently working on.
This week, I have been working on a presentation for myself to help understand and analyze my results. I looked at each Gene ontology plot by writing descriptions about each of them and analyzing what they can mean. I looked at the few important pathways and the genes that proved to be important in these plots. From there, I moved to the single-cell results, analyzing the t-SNE plots and UMAP plots by identifying similar clusters. Currently, I am going through each gene I got from the filtered differentially expressed genes (DEGs) and determining each gene's function to hopefully understand and list the ones that correspond to my research question. I have accumulated a lot of graphs and files throughout this project, so for the next few weeks, I will be working to put those pieces together to help answer my research question and determine the immunological implications of IPF.
The next few posts will be short, but I hope to see you all next week
Pragyaa
Week 12: Blog Post
Hi everyone!
This week, I was able to finish searching all the genes that I mentioned the previous week. I got quite a few pseudogenes and non-coding genes, but also found some genes that relate to immune cells and different brain cells. I have also started working on my paper by compiling all the research I have read about the lungs and IPF for my introduction, while summarizing my blog posts for my methodology, and summarizing my presentation on my results for the results section. All I need to do now is compile all the information I have, formalize the language, and then my paper will be complete.
Thank you so much for reading!!
Pragyaa
Week 13: Blog Post
Hi everyone!
This week, my focus was mainly on writing the paper and working on my presentation. I worked on my introduction, methodology, and results for my paper and my presentation. I am not entirely sure if my filtered DEGs are correct, however, I will run my code again and do some research. For some reason, a lot of my filtered DEGs are in the brain, and while some research proves a connection between the brain and IPF (I am still trying to understand those papers), I am uncertain about my results. Hopefully, all the problems will be ironed out, and I should be ready for the end of this long project.
Thank you so much for staying till the end (almost there!)
Pragyaa