Pages

Showing posts with label RNA-Seq. Show all posts
Showing posts with label RNA-Seq. Show all posts

ENCODE ChIP-Seq Significance Tool: Which TFs Regulate my Genes?

I collaborate with several investigators on gene expression projects using both microarray and RNA-seq. After I show a collaborator which genes are dysregulated in a particular condition or tissue, the most common question I get is "what are the transcription factors regulating these genes?"

This isn't the easiest question to answer. You could look at transcription factor binding site position weight matrices like those from TRANSFAC and come up with a list of all factors that potentially hit that site, then perform some kind of enrichment analysis on that. But this involves some programming, and is based solely on sequence motifs, not experimental data.

The ENCODE consortium spent over $100M and generated hundreds of ChIP-seq experiments for different transcription factors and histone modifications across many cell types (if you don't know much about ENCODE, go read the main ENCODE paper, and Sean Eddy's very fair commentary). Regardless of what you might consider "biologically functional", the ENCODE project generated a ton of data, and much of this data is publicly available. But that still doesn't help answer our question, because genes are often bound by multiple TFs, and TFs can bind many regions. We need to perform an enrichment (read: hypergeometric) test to assess an over-representation of experimentally bound transcription factors around our gene targets of interest ("around" also implies that some spatial boundary must be specified). To date, I haven't found a good tool to do this easily.

Raymond Auerbach and Bin Chen in Atul Butte's lab recently developed a resource to address this very common need, called the ENCODE ChIP-Seq Significance Tool.

The paper: Auerbach et al. Relating Genes to Function: Identifying Enriched Transcription Factors using the ENCODE ChIP-Seq Significance Tool. Bioinformatics (2013): 10.1093/bioinformatics/btt316.

The software: ENCODE ChIP-Seq Significance Tool (http://encodeqt.stanford.edu/).

This tool takes a list of "interesting" (significant, dysregulated, etc.) genes as input, and identifies ENCODE transcription factors from this list. Head over to http://encodeqt.stanford.edu/, select the ID type you're using (Ensembl, Symbol, etc), and paste in your list of genes. You can also specify your background set (this has big implications for the significance testing using the hypergeometric distribution). Scroll down some more to tell the tool how far up and downstream you want to look from the transcription start/end site or whole gene, select an ENCODE cell line (or ALL), and hit submit. 

You're then presented with a list of transcription factors that are most likely regulating your input genes (based on overrepresentation of ENCODE ChIP-seq binding sites). Your results can then be saved to CSV or PDF. You can also click on a number in the results table and get a list of genes that are regulated by a particular factor (the numbers do not appear as hyperlinks in my browser, but clicking the number still worked).

At the very bottom of the page, you can load example data that they used in the supplement of their paper, and run through the analysis presented therein. The lead author, Raymond Auerbach, even made a very informative screencast on how to use the tool:


Now, if I could only find a way to do something like this with mouse gene expression data.

Automated Archival and Visual Analysis of Tweets Mentioning #bog13, Bioinformatics, #rstats, and Others

Automatically Archiving Twitter Results

Ever since Twitter gamed its own API and killed off great services like IFTTT triggers, I've been looking for a way to automatically archive tweets containing certain search terms of interest to me. Twitter's built-in search is limited, and I wanted to archive interesting tweets for future reference and to start playing around with some basic text / trend analysis.

Enter t - the twitter command-line interface. t is a command-line power tool for doing all sorts of powerful Twitter queries using the command line. See t's documentation for examples.

I wrote this script that uses the t utility to search Twitter separately for a set of specified keywords, and append those results to a file. The comments at the end of the script also show you how to commit changes to a git repository, push to GitHub, and automate the entire process to run twice a day with a cron job. Here's the code as of May 14, 2013:



That script, and results for searching for "bioinformatics", "metagenomics", "#rstats", "rna-seq", and "#bog13" (the Biology of Genomes 2013 meeting) are all in the GitHub repository below. (Please note that these results update dynamically, and searching Twitter at any point could possibly result in returning some unsavory Tweets.)

https://github.com/stephenturner/twitterchive

Analyzing Tweets using R

You'll also find an analysis subdirectory, containing some R code to produce barplots showing the number of tweets per day over the last month, frequency of tweets by hour of the day, the most used hashtags within a search, the most prolific tweeters, and a ubiquitous word cloud. Much of this code is inspired by Neil Saunders's analysis of Tweets from ISMB 2012. Here's the code as of May 14, 2013:



Also in that analysis directory you'll see periodically updated plots for the results of the queries above.

Analyzing Tweets mentioning "bioinformatics"

Using the bioinformatics query, here are the number of tweets per day over the last month:


Here is the frequency of "bioinformatics" tweets by hour:

Here are the most used hashtags (other than #bioinformatics):

Here are the most prolific bioinformatics Tweeps:

Here's a wordcloud for all the bioinformatics Tweets since March:

Analyzing Tweets mentioning "#bog13"

The 2013 CSHL Biology of Genomes Meeting took place May 7-11, 2013. I searched and archived Tweets mentioning #bog13 from May 1 through May 14 using this script. You'll notice in the code above that I'm no longer archiving this hashtag. I probably need a better way to temporarily add keywords to the search, but I haven't gotten there yet.

Here are the number of Tweets per day during that period. Tweets clearly peaked a couple days into the meeting, with follow-up commentary trailing off quickly after the meeting ended.


Here is the frequency frequency of Tweets by hour, clearly bimodal:

Top hashtags (other than #bog13). Interestingly #bog14 was the most highly used hashtag, so I'm guessing lots of folks are looking forward to next years' meeting. Also, #ashg12 got lots of mentions, presumably because someone presented updated work from last years' ASHG meeting.

Here were the most prolific Tweeps - many of the usual suspects here, as well as a few new ones (new to me at least):

And finally, the requisite wordcloud:


More analysis

If you look in the analysis directory of the repo you'll find plots like these for other keywords (#rstats, metagenomics, rna-seq, and others to come). I would also like to do some sentiment analysis as Neil did in the ISMB post referenced above, but the sentiment package has since been removed from CRAN. I hear there are other packages for polarity analysis, but I haven't yet figured out how to use them. I've given you the code to do the mundane stuff (parsing the fixed-width files from t, for starters). I'd love to see someone take a stab at some further text mining / polarity / sentiment analysis!

twitterchive - archive and analyze results from a Twitter search

Scotty, We Need More Power! Power, Sample Size, and Coverage Estimation for RNA-Seq

Two of the most common questions at the beginning of an RNA-seq experiments are "how many reads do I need?" and "how many replicates do I need?". This paper describes a web application for designing RNA-seq applications that calculates an appropriate sample size and read depth to satisfy user-defined criteria such as cost, maximum number of reads or replicates attainable, etc. The power and sample size estimations are based on a t-test, which the authors claim, performs no worse than the negative binomial models implemented by popular RNA-seq methods such as DESeq, when there are three or more replicates present. Empirical distributions are taken from either (1) pilot data that the user can upload, or (2) built in publicly available data. The authors find that there is substantial heterogeneity between experiments (technical variation is larger than biological variation in many cases), and that power and sample size estimation will be more accurate when the user provides their own pilot data.

My only complaint, for all the reasons expressed in my previous blog post about why you shouldn't host things like this exclusively on your lab website, is that the code to run this analysis doesn't appear to be available to save, study, modify, maintain, or archive. When lead author Michele Busby leaves Gabor Marth's lab, hopefully the app doesn't fall into the graveyard of computational biology web apps.  Update 2/7/13: Michele Busby created a public Github repository for the Scotty code: https://github.com/mbusby/Scotty

tl;dr? There's a new web app that does power, sample size, and coverage calculations for RNA-seq, but it only works well if the pilot or public data you give it closely matches the actual data you'll collect. 



Differential Isoform Expression With RNA-Seq: Are We Really There Yet?

In case you missed it, a new paper was published in Nature Biotechnology on a method for detecting isoform-level differential expression with RNA-seq Data:

Trapnell, Cole, et al. "Differential analysis of gene regulation at transcript resolution with RNA-seq." Nature Biotechnology (2012).

THE PROBLEM

RNA-seq enables transcript-level resolution of gene expression, but there is no proven methodology for simultaneously accounting for biological variability across replicates and uncertainty in mapping fragments to isoforms. One of the most commonly used workflows is to map reads with a tool like Tophat or STAR, use a tool like HTSeq to count the number of reads overlapping a gene, then use a negative-binomial count-based approach such as edgeR or DESeq to assess differential expression at the gene level.  

Figure 1 in the paper illustrates the problem with existing approaches, which only count the number of fragments originating from either the entire gene or constitutive exons only. 

Excerpt from figure 1 from the Cuffdiff 2 paper.

In the top row, a change in gene expression is undetectable by counting reads mapping to any exon, and is underestimated if counting only constitutive exons. In the middle row, an apparent change would be detected, but in the wrong direction if using a count-based method alone rather than accounting for which transcript a read comes from and how long that transcript is. How often situations like the middle row happen in reality, that's anyone's guess.

THE PROPOSED SOLUTION

The method presented in this paper, popularized by the cuffdiff method in the Cufflinks software package, claims to address both of these problems simultaneously by modeling variability in the number of fragments generated by each transcript across biological replicates using a beta negative binomial mixture distribution that accounts for both sources of variability in a transcript's measured expression level. This so-called transcript deconvolution is not computationally trivial, and incredibly difficult to explain, but failure to account for the uncertainty (measurement error) from which transcript a fragment originates from can result in a high false-positive rate, especially when there is significant differential regulation of isoforms. Compared to existing methods, the procedure described claims equivalent sensitivity with a much lower false-positive rate when there is substantial isoform-level variability in gene expression between conditions. 

ALTERNATIVE WORKFLOWS

Importantly, the manuscript also addresses and points out weaknesses several undocumented "alternative" workflows that are discussed often on forums like SEQanswers and anecdotally at meetings. These alternative workflows are variations on a theme: combining transcript-level fragment count estimates (like estimates from Cufflinks, eXpress, or RSEM mapping to a transcriptome), with downstream count-based analysis tools like edgeR/DESeq (both R/Bioconductor packages). This paper points out that none of these tools were meant to be used this way, and doing so violates assumptions of underlying statistics used by  both procedures. However, the authors concede that the variance modeling strategies of edgeR and DESeq are robust, and thus assessed the performance of these "alternative" workflows. The results of those experiments show that the algorithm presented in this paper, cuffdiff 2, outperforms other alternative hybrid Cufflinks/RSEM + edgeR/DESeq workflows [see supplementary figure 77 (yes, 77!]).

REPRODUCIBILITY ISSUES

In theory (and in the simulation studies presented here, see further comments below), the methodology presented here seems to outperform any other competing workflow. So why isn't everyone using it, and why is there so much grumbling about it on forums and at meetings? For many (myself included), the biggest issue is one of reproducibility. There are many discussions about cufflinks/cuffdiff providing drastically different results from one version to the next (see here, here, here, here, and here, for a start). The core I run operates in a production environment where everything I do must be absolutely transparent and reproducible. Reporting drastically different results to my collaborators whenever I update the tools I'm using is very alarming to a biologist, and reduces their confidence in the service I provide and the tools I use. 

Furthermore, a recent methods paper recently compared their tool, DEXSeq, to several different versions of cuffdiff. Here, the authors performed two comparisons: a "proper" comparison, where replicates of treatments (T1-T3) were compared to replicates of controls (C1-C4), and a "mock" comparison, where controls (e.g. C1+C3) were compared to other controls (C2+C4). The most haunting result is shown below, where the "proper" comparison finds relatively few differentially expressed genes, while the "mock" comparison of controls versus other controls finds many, many more differentially expressed genes, and an increasing number with newer versions of cufflinks:

Table S1 from the DEXSeq paper.
This comparison predates the release of Cuffdiff 2, so perhaps this alarming trend ceases with the newer release of Cufflinks. However, it is worth noting that these data shown here are from a real dataset, where all the comparisons in the new Cuffdiff 2 paper were done with simulations. Having done some method development myself, I realize how easy it is to construct a simulation scenario to support nearly any claim you'd like to make.

FINAL THOUGHTS

Most RNA-seq folks would say that the field has a good handle on differential expression at the gene level, while differential expression at isoform-level resolution is still under development. I would tend to agree with this statement, but if cases as presented in Figure 1 of this paper are biologically important and widespread (they very well may be), then perhaps we have some re-thinking to do, even with what we thought were "simple" analyses at the gene level. 

What's your workflow for RNA-seq analysis? Discuss.

STAR: ultrafast universal RNA-seq aligner

There's a new kid on the block for RNA-seq alignment.

Dobin, Alexander, et al. "STAR: ultrafast universal RNA-seq aligner." Bioinformatics (2012).

Aligning RNA-seq data is challenging because reads can overlap splice junctions. Many other RNA-seq alignment algorithms (e.g. Tophat) are built on top of DNA sequence aligners. STAR (Spliced Transcripts Alignment to a Reference) is a standalone RNA-seq alignment algorithm that uses uncompressed suffix arrays and a mapping algorithm similar to those used in large-scale genome alignment tools to align RNA-seq reads to a genomic reference. STAR is over 50 times faster than any other previously published RNA-seq aligner, and outperforms other aligners in both sensitivity and specificity using both simulated and real (replicated) RNA-seq data.



The notable increase in speed comes at the price of a larger memory requirement. STAR requires ~27GB RAM to align reads to a human genome - a moderate amount, but not atypical on most modern servers. STAR aligns ~45 million paired reads per hour per processor, and scales nearly linearly with the number of processors (without appreciably increasing RAM usage). Notably, the STAR algorithm is also capable of handling longer reads such as those from PacBio and the upcoming Oxford Nanopore technologies. STAR is free and open source software.

Dobin, Alexander, et al. "STAR: ultrafast universal RNA-seq aligner." Bioinformatics (2012).

STAR software on Google Code

(This post adapted from my review on F1000).

DESeq vs edgeR Comparison

Update (Dec 18, 2012): Please see this related post I wrote about differential isoform expression analysis with Cuffdiff 2.

DESeq and edgeR are two methods and R packages for analyzing quantitative readouts (in the form of counts) from high-throughput experiments such as RNA-seq or ChIP-seq. After alignment, reads are assigned to a feature, where each feature represents a target transcript, in the case of RNA-Seq, or a binding region, in the case of ChIP-Seq. An important summary statistic is the count of the number of reads in a feature (for RNA-Seq, this read count is a good approximation of transcript abundance).

Methods used to analyze array-based data assume a normally distributed, continuous response variable. However, response variables for digital methods like RNA-seq and ChIP-seq are discrete counts. Thus, both DESeq and edgeR methods are based on the negative binomial distribution.

I see these two tools often used interchangeably, and I wanted to take a look at how they stack up to one another in terms of performance, ease of use, and speed. This isn't meant to be a comprehensive evaluation or "bake-off" between the two methods. This would require complex simulations, parameter sweeps, and evaluation with multiple well-characterized real RNA-seq datasets. Further, this is only a start - a full evaluation would need to be much more comprehensive.

Here, I used the newest versions of both edgeR and DESeq, using the well-characterized Pasilla dataset, available in the pasilla Bioconductor package. The dataset is from an experiment in Drosophila investigating the effect of RNAi knockdown of the splicing factor, pasilla. I used the GLM functionality of both packages, as recommended by the vignettes, for dealing with a multifactorial experiment (condition: treated vs. untreated; library type: single-end and paired-end).



Both packages provide built-in functions for assessing overall similarity between samples using either PCA (DESeq) or MDS (edgeR), although these methods operate on the same underlying data and could easily be switched.

PCA plot on variance stabilized data from DESeq:

MDS plot from edgeR:


Per gene dispersion estimates from DESeq:

Biological coefficient of variation versus abundance (edgeR):


Now, let's see how many statistically significant (FDR<0.05) results each method returns:



In this simple example, DESeq finds 820 genes significantly differentially expressed at FDR<0.05, while edgeR is finds these 820 and an additional 371. Let's take a look at the detected fold changes from both methods:

Here, if genes were found differentially expressed by edgeR only, they're colored red; if found by both, colored green. What's striking here is that for a handful of genes, DESeq is (1) reporting massive fold changes, and (2) not calling them statistically significant. What's going on here?

It turns out that these genes have extremely low counts (usually one or two counts in only one or two samples). The DESeq vignette goes through the logic of independent filtering, showing that the likelihood of a gene being significantly differentially expressed is related to how strongly it's expressed, and advocates for discarding extremely lowly expressed genes, because differential expression is likely not statistically detectable.

Count-based filtering can be achieved two ways. The DESeq vignette demonstrates how to filter based on quantiles, while I used the filtering method demonstrated in the edgeR vignette - removing genes without at least 2 counts per million in at least two samples. This filtering code is commented out above - uncomment to filter.

After filtering, all of the genes shown above with apparently large fold changes as detected by DESeq are removed prior to filtering, and the fold changes correlate much better between the two methods. edgeR still detects ~50% more differentially expressed genes, and it's unclear to me (1) why this is the case, and (2) if this is necessarily a good thing.


Conclusions:

Unfortunately, I may have oversold the title here - this is such a cursory comparison of the two methods that I would hesitate to draw any conclusions about which method is better than the other. In addition to finding more significantly differentially expressed genes (again, not necessarily a good thing), I can say that edgeR was much faster than DESeq for fitting GLM models, but it took slightly longer to estimate the dispersion. Further without any independent filtering, edgeR gave me moderated fold changes for the extremely lowly expressed genes for which DESeq returned logFCs in the 20-30 range (but these transcripts were so lowly expressed anyway, they should have been filtered out before any evaluation).

If there's one thing that will make me use edgeR over DESeq (until I have time to do a more thorough evaluation), it's the fact that using edgeR seems much more natural than DESeq, especially if you're familiar with the limma package (pretty much the standard for analyzing microarray data and other continuously distributed gene expression data). Setting up the design matrix and specifying contrasts feels natural if you're familiar with using limma. Further, the edgeR user guide weighs in at 67 pages, filled with many case studies that will help you in putting together a design matrix for nearly any experimental design: paired designs, time courses, batch effects, interactions, etc. The DESeq documentation is still fantastic, but could benefit from a few more case studies / examples.

What do you think? Anyone want to fork my R code and help do this comparison more comprehensively (more examples, simulated data, speed benchmarking)? Is the analysis above fair? What do you find more easy to use, or is ease-of-use (and thus, reproducibility) even important when considering data analysis?

RNA-Seq & ChiP-Seq Data Analysis Course at EBI

I just got this announcement from EMBL-EBI about an RNA-seq/ChIP-seq analysis hands-on course. Find the full details, schedule, and speaker list here.

Title: Advanced RNA-Seq and Chip-Seq Data Analysis Course
Date: May 1-4 2012
Venue: EMBL-EBI, Hinxton, Nr Cambridge, CB10 1SD, UK
Registration Closing Date: March 6 2012 (12:00 midday GMT)

This course is aimed at advanced PhD students and post-doctoral researchers who are applying or planning to apply high throughput sequencing technologies and bioinformatics methods in their research. The aim of this course is to familiarize the participants with advanced data analysis methodologies and provide hands-on training on the latest analytical approaches.

Lectures will give insight into how biological knowledge can be generated from RNA-seq and ChIP-seq experiments and illustrate different ways of analyzing such data Practicals will consist of computer exercises that will enable the participants to apply statistical methods to the analysis of RNA-seq and ChIP-seq data under the guidance of the lecturers and teaching assistants. Familiarity with the technology and biological use cases of high throughput sequencing is required, as is some experience with R/Bioconductor.

The course covers data analysis of RNA-Seq and ChIP-Seq experiments.
Topics will include: alignment, data handling and visualisation, region identification, differential expression, data quality assessment and statistical analysis, using R/Bioconductor.

An example RNA-Seq Quality Control and Analysis Workflow

I found the slides below on the education page from Bioinformatics & Research Computing at the Whitehead Institute. The first set (PDF) gives an overview of the methods and software available for quality assessment of microarray and RNA-seq experiments using the FastX toolkit and FastQC.



The second set (PDF)  gives an example RNA-seq workflow using TopHat, SAMtools, Python/HTseq, and R/DEseq.



If you're doing any RNA-seq work these are both really nice resources to help you get a command-line based analysis workflow up and running (if you're not using Galaxy for RNA-seq).

Guide to RNA-seq Analysis in Galaxy

James Taylor came to UVA last week and gave an excellent talk on how Galaxy enables transparent and reproducible research in genomics. I'm gearing up to take on several projects that involve next-generation sequencing, and I'm considering installing my own Galaxy framework on a local cluster or on the cloud.

If you've used Galaxy in the past you're probably aware that it allows you to share data, workflows, and histories with other users. New to me was the pages section, where an entire analysis is packaged on a single pages, and vetting is crowdsourced to other Galaxy users in the form of comments and voting.

I recently found a page published by Galaxy user Jeremy that serves as a guide to RNA-seq analysis using Galaxy. If you've never done RNA-seq before it's a great place to start. The guide has all the data you need to get started on an experiment where you'll use TopHat/Bowtie to align reads to a reference genome, and Cufflinks to assemble transcripts and quantify differential gene expression, alternative splicing, etc. The dataset is small, so all the analyses start and finish quickly, allowing you to finish the tutorial in just a few hours. The author was kind enough to include links to relevant sections of the TopHat and Cufflinks documentation where it's needed in the tutorial. Hit the link below to get started.

Galaxy Pages: RNA-seq Analysis Exercise