metaseqR report

Summary

Analysis summary

Summary: The raw bam files, one for each RNA-Seq sample, were summarized to a gene read counts table, using the Bioconductor package GenomicRanges. In the final read counts table, each row represented one gene, each column one RNA-Seq sample and each cell, the corresponding read counts associated with each row and column.The gene counts table was normalized for inherent systematic or experimental biases (e.g. sequencing depth, gene length, GC content bias etc.) using the Bioconductor package edgeR after removing genes that had zero counts over all the RNA-Seq samples (1338 genes). The output of the normalization algorithm was a table with normalized counts, which can be used for differential expression analysis with statistical algorithms developed specifically for count data. Prior to the statistical testing procedure, the gene read counts were filtered for possible artifacts that could affect the subsequent statistical testing procedures. Genes presenting any of the following were excluded from further analysis: i) genes with length less than 500 (175 genes), ii) genes whose average reads per 100 bp was less than the 25^th quantile of the total normalized distribution of average reads per 100bp (0 genes with cutoff value 0.12634 average reads per 100 bp), iii) genes with read counts below the median read counts of the total normalized count distribution (9585 genes with cutoff value 442 normalized read counts). The total number of genes excluded due to the application of gene filters was 4524. The total (unified) number of genes excluded due to the application of all filters was 10965. The resulting gene counts table was subjected to differential expression analysis for the contrasts WT versus TR using the Bioconductor package edgeR. The final numbers of differentially expressed genes were (per contrast): for the contrast WT versus TR, 212 (70) statistically significant genes were found with a p-value (FDR or adjusted p-value) threshold of 0.05 and of these, 33 (5) were up-regulated, 105 (65) were down-regulated and 74 (0) were not differentially expressed according to an absolute fold change cutoff value of 1 in log₂ scale. Literature references for all the algorithms used can be found at the end of this report.

Input options

Read counts file: imported sam/bam/bed files
Conditions: TR, WT
Samples: SCR1, SCR2
Requested contrasts: WT_vs_TR
Library sizes:

SCR1: 38884842
SCR2: 33471113

Annotation: download
Organism: mouse (Mus musculus), genome version alias mm9
Annotation source: RefSeq database
Count type: gene
Exon filters: min.active.exons

min.active.exons
- exons.per.gene: 5
- min.exons: 2
- frac: 0.2

Gene filters: length, avg.reads, expression, biotype

length
- length: 500

avg.reads
- average.per.bp: 100
- quantile: 0.25

expression
- median: TRUE
- mean: FALSE
- quantile: NA
- known: NA
- custom: NA

biotype
- pseudogene: FALSE
- snRNA: FALSE
- protein_coding: FALSE
- antisense: FALSE
- miRNA: FALSE
- lincRNA: FALSE
- snoRNA: FALSE
- processed_transcript: FALSE
- misc_RNA: FALSE
- rRNA: TRUE
- sense_overlapping: FALSE
- sense_intronic: FALSE
- polymorphic_pseudogene: FALSE
- non_coding: FALSE
- three_prime_overlapping_ncrna: FALSE
- IG_C_gene: FALSE
- IG_J_gene: FALSE
- IG_D_gene: FALSE
- IG_V_gene: FALSE
- ncrna_host: FALSE

Filter application: after normalization
Normalization algorithm: edgeR
Normalization arguments: method, refColumn, logratioTrim, sumTrim, doWeighting, Acutoff, p

method: TMM
logratioTrim: 0.3
sumTrim: 0.05
doWeighting: TRUE
Acutoff: -1e+10
p: 0.75

Statistical algorithm(s): edgeR
Statistical arguments for edgeR: main.method, rowsum.filter, prior.df, trend, span, tag.method, grid.length, grid.range, offset, glm.method, subset, AveLogCPM, trend.method, dispersion, offset, weights, lib.size, prior.count, start, method, test, abundance.trend, robust, winsor.tail.p

main.method: classic
rowsum.filter: 5
prior.df: 10
trend: movingave
tag.method: grid
grid.length: 11
grid.range: -6, 6
glm.method: CoxReid
subset: 10000
trend.method: auto
prior.count: 0.125
method: auto
test: chisq
abundance.trend: TRUE
robust: FALSE
winsor.tail.p: 0.05, 0.1

Meta-analysis method: no meta-analysis, reported p-values from the first supplied statistical algorithm
Multiple testing correction: Benjamini-Hochberg FDR
p-value threshold: 0.05
Logarithmic tranformation offset: 1
Analysis preset: not available
Quality control plots: multidimensional scaling, biotype detection, biotype counts, sample and biotype saturation, correlation heatmap and correlogram, pairwise scatterplots between samples, boxplots, volcano plot, DEG biotype detection, DEG heatmap
Figure format: png, pdf
Output directory: /data/images/proton/Cereghini/run142/metaseqr_wt_vs_tr_run20
Output data: Annotation, p-value, Adjusted p-value (FDR), Fold change, Statistics, Read counts
Output scale(s): log2 scale
Output values: Normalized values
Output statistics: Mean, Median, Standard deviation, Median Absolute Deviation (MAD), Coefficient of Variation, Robust Coefficient of Variation
Total run time: 15 minutes 09 seconds

Filtered genes

Number of filtered genes: 10965 which is the union of

Filtered because of zero reads: 1338
Filtered because of exon filters: 0
Filtered because of gene filters: 9627 which is the union of
- length: 175 genes with filter cutoff value 500
- avg.reads: 4346 genes with filter cutoff value 0.1263387
- expression: 9585 genes further decomposed to (filter name, filtered genes, filter cutoff):
- - median: 9585 genes with filter cutoff value 442
- biotype: 2 genes with filter cutoff value rRNA

Differentially expressed genes

Number of differentially expressed genes per contrast:

WT_vs_TR: 212 (70) statistically significant genes of which 33 (5) up regulated, 105 (65) down regulated and 74 (0) not differentially expressed according to a p-value (FDR or adjusted p-value) threshold of 0.05 and an absolute fold change cutoff value of 1 in log₂ scale.

Run log

[ 2015-02-25 12:54:46 ] INFO 2015-02-25 12:54:46: Data processing started...
[ 2015-02-25 12:54:46 ] INFO Read counts file: imported sam/bam/bed files
[ 2015-02-25 12:54:46 ] INFO Conditions: TR, WT
[ 2015-02-25 12:54:46 ] INFO Samples: SCR1, SCR2
[ 2015-02-25 12:54:46 ] INFO Requested contrasts: WT_vs_TR
[ 2015-02-25 12:54:46 ] INFO Annotation: download
[ 2015-02-25 12:54:46 ] INFO Organism: mm9
[ 2015-02-25 12:54:46 ] INFO Reference source: refseq
[ 2015-02-25 12:54:46 ] INFO Count type: gene
[ 2015-02-25 12:54:46 ] INFO Exon filters: min.active.exons
[ 2015-02-25 12:54:46 ] INFO min.active.exons:
[ 2015-02-25 12:54:46 ] INFO exons.per.gene: 5
[ 2015-02-25 12:54:46 ] INFO min.exons: 2
[ 2015-02-25 12:54:46 ] INFO frac: 0.2
[ 2015-02-25 12:54:46 ] INFO Gene filters: length, avg.reads, expression, biotype
[ 2015-02-25 12:54:46 ] INFO length:
[ 2015-02-25 12:54:46 ] INFO length: 500
[ 2015-02-25 12:54:46 ] INFO avg.reads:
[ 2015-02-25 12:54:46 ] INFO average.per.bp: 100
[ 2015-02-25 12:54:46 ] INFO quantile: 0.25
[ 2015-02-25 12:54:46 ] INFO expression:
[ 2015-02-25 12:54:46 ] INFO median: TRUE
[ 2015-02-25 12:54:46 ] INFO mean: FALSE
[ 2015-02-25 12:54:46 ] INFO quantile: NA
[ 2015-02-25 12:54:46 ] INFO known: NA
[ 2015-02-25 12:54:46 ] INFO custom: NA
[ 2015-02-25 12:54:46 ] INFO biotype:
[ 2015-02-25 12:54:46 ] INFO pseudogene: FALSE
[ 2015-02-25 12:54:46 ] INFO snRNA: FALSE
[ 2015-02-25 12:54:46 ] INFO protein_coding: FALSE
[ 2015-02-25 12:54:46 ] INFO antisense: FALSE
[ 2015-02-25 12:54:46 ] INFO miRNA: FALSE
[ 2015-02-25 12:54:46 ] INFO lincRNA: FALSE
[ 2015-02-25 12:54:46 ] INFO snoRNA: FALSE
[ 2015-02-25 12:54:46 ] INFO processed_transcript: FALSE
[ 2015-02-25 12:54:46 ] INFO misc_RNA: FALSE
[ 2015-02-25 12:54:46 ] INFO rRNA: TRUE
[ 2015-02-25 12:54:46 ] INFO sense_overlapping: FALSE
[ 2015-02-25 12:54:46 ] INFO sense_intronic: FALSE
[ 2015-02-25 12:54:46 ] INFO polymorphic_pseudogene: FALSE
[ 2015-02-25 12:54:46 ] INFO non_coding: FALSE
[ 2015-02-25 12:54:46 ] INFO three_prime_overlapping_ncrna: FALSE
[ 2015-02-25 12:54:46 ] INFO IG_C_gene: FALSE
[ 2015-02-25 12:54:46 ] INFO IG_J_gene: FALSE
[ 2015-02-25 12:54:46 ] INFO IG_D_gene: FALSE
[ 2015-02-25 12:54:46 ] INFO IG_V_gene: FALSE
[ 2015-02-25 12:54:46 ] INFO ncrna_host: FALSE
[ 2015-02-25 12:54:46 ] INFO Filter application: postnorm
[ 2015-02-25 12:54:46 ] INFO Normalization algorithm: edger
[ 2015-02-25 12:54:46 ] INFO Normalization arguments:
[ 2015-02-25 12:54:46 ] INFO method: TMM
[ 2015-02-25 12:54:46 ] INFO logratioTrim: 0.3
[ 2015-02-25 12:54:46 ] INFO sumTrim: 0.05
[ 2015-02-25 12:54:46 ] INFO doWeighting: TRUE
[ 2015-02-25 12:54:46 ] INFO Acutoff: -1e+10
[ 2015-02-25 12:54:46 ] INFO p: 0.75
[ 2015-02-25 12:54:46 ] INFO Statistical algorithm: edger
[ 2015-02-25 12:54:46 ] INFO Statistical arguments:
[ 2015-02-25 12:54:46 ] INFO edger: classic, 5, 10, movingave, NULL, grid, 11, c(-6, 6), NULL, CoxReid, 10000, NULL, auto, NULL, NULL, NULL, NULL, 0.125, NULL, auto, chisq, TRUE, FALSE, c(0.05, 0.1)
[ 2015-02-25 12:54:46 ] INFO Meta-analysis method: none
[ 2015-02-25 12:54:46 ] INFO Multiple testing correction: BH
[ 2015-02-25 12:54:46 ] INFO p-value threshold: 0.05
[ 2015-02-25 12:54:46 ] INFO Logarithmic transformation offset: 1
[ 2015-02-25 12:54:46 ] INFO Quality control plots: mds, biodetection, countsbio, saturation, readnoise, correl, pairwise, boxplot, volcano, biodist, deheatmap
[ 2015-02-25 12:54:46 ] INFO Figure format: png, pdf
[ 2015-02-25 12:54:46 ] INFO Output directory: /data/images/proton/Cereghini/run142/metaseqr_wt_vs_tr_run20
[ 2015-02-25 12:54:46 ] INFO Output data: annotation, p.value, adj.p.value, fold.change, stats, counts
[ 2015-02-25 12:54:46 ] INFO Output scale(s): log2
[ 2015-02-25 12:54:46 ] INFO Output values: normalized
[ 2015-02-25 12:54:46 ] INFO Output statistics: mean, median, sd, mad, cv, rcv
[ 2015-02-25 12:54:46 ] INFO Downloading gene annotation for mm9...
[ 2015-02-25 12:55:01 ] INFO Converting annotation to GenomicRanges object...
[ 2015-02-25 12:55:01 ] INFO Getting DNA sequences...
[ 2015-02-25 12:55:10 ] INFO Getting GC content...
[ 2015-02-25 12:55:11 ] INFO Converting annotation to GenomicRanges object...
[ 2015-02-25 12:55:11 ] INFO Reading bam file sort_uniq.bam for sample with name SCR1. This might take some time...
[ 2015-02-25 12:55:12 ] INFO Reading bam file sort_uniq.bam for sample with name SCR2. This might take some time...
[ 2015-02-25 12:57:26 ] INFO Counting reads overlapping with given annotation...
[ 2015-02-25 12:57:26 ] INFO ...for single-end reads...
[ 2015-02-25 12:57:26 ] INFO ...using strandedness...
[ 2015-02-25 12:57:36 ] INFO Counting reads overlapping with given annotation...
[ 2015-02-25 12:57:36 ] INFO ...for single-end reads...
[ 2015-02-25 12:57:36 ] INFO ...using strandedness...
[ 2015-02-25 13:08:25 ] INFO Exporting raw read counts table to /data/images/proton/Cereghini/run142/metaseqr_wt_vs_tr_run20/lists/raw_counts_table.txt.gz
[ 2015-02-25 13:08:26 ] INFO Saving gene model to /data/images/proton/Cereghini/run142/metaseqr_wt_vs_tr_run20/data/gene_model.RData
[ 2015-02-25 13:08:26 ] INFO Removing genes with zero counts in all samples...
[ 2015-02-25 13:08:26 ] INFO Normalizing with: edger
[ 2015-02-25 13:08:26 ] INFO Applying gene filter length...
[ 2015-02-25 13:08:26 ] INFO Threshold below which ignored: 500
[ 2015-02-25 13:08:26 ] INFO Applying gene filter avg.reads...
[ 2015-02-25 13:08:26 ] INFO Threshold below which ignored: 0.126338748030583
[ 2015-02-25 13:08:26 ] INFO Applying gene filter expression...
[ 2015-02-25 13:08:26 ] INFO Threshold below which ignored: 442
[ 2015-02-25 13:08:26 ] INFO Applying gene filter biotype...
[ 2015-02-25 13:08:26 ] INFO Biotypes ignored: rRNA
[ 2015-02-25 13:08:26 ] INFO 10965 genes filtered out
[ 2015-02-25 13:08:26 ] INFO 10129 genes remain after filtering
[ 2015-02-25 13:08:26 ] INFO Running statistical tests with: edger
[ 2015-02-25 13:08:26 ] WARN No replication when testing with edgeR! Consider using another statistical test or just performing empirical analysis. Setting to 0.2...
[ 2015-02-25 13:08:26 ] INFO Contrast: WT_vs_TR
[ 2015-02-25 13:08:28 ] INFO Contrast WT_vs_TR: found 212 genes
[ 2015-02-25 13:08:28 ] INFO Exporting and compressing normalized read counts table to /data/images/proton/Cereghini/run142/metaseqr_wt_vs_tr_run20/lists/normalized_counts_table.txt
[ 2015-02-25 13:08:29 ] INFO Building output files...
[ 2015-02-25 13:08:29 ] INFO Contrast: WT_vs_TR
[ 2015-02-25 13:08:29 ] INFO Adding non-filtered data...
[ 2015-02-25 13:08:29 ] INFO binding annotation...
[ 2015-02-25 13:08:29 ] INFO binding p-values...
[ 2015-02-25 13:08:29 ] INFO binding FDRs...
[ 2015-02-25 13:08:30 ] INFO binding log2 normalized fold changes...
[ 2015-02-25 13:08:30 ] INFO binding normalized mean counts...
[ 2015-02-25 13:08:30 ] INFO binding normalized median counts...
[ 2015-02-25 13:08:31 ] INFO binding normalized count sds...
[ 2015-02-25 13:08:32 ] INFO binding normalized count MADs...
[ 2015-02-25 13:08:33 ] INFO binding normalized count CVs...
[ 2015-02-25 13:08:34 ] INFO binding normalized counts RCVs...
[ 2015-02-25 13:08:36 ] INFO binding normalized mean counts...
[ 2015-02-25 13:08:36 ] INFO binding normalized median counts...
[ 2015-02-25 13:08:37 ] INFO binding normalized count sds...
[ 2015-02-25 13:08:37 ] INFO binding normalized count MADs...
[ 2015-02-25 13:08:39 ] INFO binding normalized count CVs...
[ 2015-02-25 13:08:40 ] INFO binding normalized counts RCVs...
[ 2015-02-25 13:08:41 ] INFO binding all normalized counts for WT...
[ 2015-02-25 13:08:42 ] INFO binding all normalized counts for TR...
[ 2015-02-25 13:08:42 ] INFO Writing output...
[ 2015-02-25 13:08:42 ] INFO Adding filtered data...
[ 2015-02-25 13:08:42 ] INFO binding annotation...
[ 2015-02-25 13:08:42 ] INFO binding p-values...
[ 2015-02-25 13:08:42 ] INFO binding FDRs...
[ 2015-02-25 13:08:42 ] INFO binding log2 normalized fold changes...
[ 2015-02-25 13:08:42 ] INFO binding normalized mean counts...
[ 2015-02-25 13:08:42 ] INFO binding normalized median counts...
[ 2015-02-25 13:08:43 ] INFO binding normalized count sds...
[ 2015-02-25 13:08:43 ] INFO binding normalized count MADs...
[ 2015-02-25 13:08:45 ] INFO binding normalized count CVs...
[ 2015-02-25 13:08:46 ] INFO binding normalized counts RCVs...
[ 2015-02-25 13:08:47 ] INFO binding normalized mean counts...
[ 2015-02-25 13:08:48 ] INFO binding normalized median counts...
[ 2015-02-25 13:08:48 ] INFO binding normalized count sds...
[ 2015-02-25 13:08:49 ] INFO binding normalized count MADs...
[ 2015-02-25 13:08:50 ] INFO binding normalized count CVs...
[ 2015-02-25 13:08:51 ] INFO binding normalized counts RCVs...
[ 2015-02-25 13:08:53 ] INFO binding all normalized counts for WT...
[ 2015-02-25 13:08:53 ] INFO binding all normalized counts for TR...
[ 2015-02-25 13:08:53 ] INFO Writing output...
[ 2015-02-25 13:08:54 ] INFO Creating quality control graphs...
[ 2015-02-25 13:08:54 ] INFO Plotting in png format...
[ 2015-02-25 13:08:54 ] INFO Plotting mds...
[ 2015-02-25 13:08:54 ] WARN MDS plot cannot be created with less than 3 samples! Skipping...
[ 2015-02-25 13:08:54 ] INFO Plotting biodetection...
[ 2015-02-25 13:08:55 ] INFO Plotting countsbio...
[ 2015-02-25 13:08:56 ] INFO Plotting saturation...
[ 2015-02-25 13:08:58 ] INFO Plotting readnoise...
[ 2015-02-25 13:08:58 ] INFO Plotting correl...
[ 2015-02-25 13:08:59 ] INFO Plotting pairwise...
[ 2015-02-25 13:09:01 ] INFO Plotting boxplot...
[ 2015-02-25 13:09:01 ] INFO Plotting boxplot...
[ 2015-02-25 13:09:01 ] INFO Plotting volcano...
[ 2015-02-25 13:09:01 ] INFO Contrast: WT_vs_TR
[ 2015-02-25 13:09:02 ] INFO Plotting biodist...
[ 2015-02-25 13:09:02 ] INFO Contrast: WT_vs_TR
[ 2015-02-25 13:09:02 ] INFO Plotting deheatmap...
[ 2015-02-25 13:09:02 ] INFO Contrast: WT_vs_TR
[ 2015-02-25 13:09:27 ] INFO Plotting in pdf format...
[ 2015-02-25 13:09:27 ] INFO Plotting mds...
[ 2015-02-25 13:09:27 ] WARN MDS plot cannot be created with less than 3 samples! Skipping...
[ 2015-02-25 13:09:27 ] INFO Plotting biodetection...
[ 2015-02-25 13:09:28 ] INFO Plotting countsbio...
[ 2015-02-25 13:09:29 ] INFO Plotting saturation...
[ 2015-02-25 13:09:30 ] INFO Plotting readnoise...
[ 2015-02-25 13:09:31 ] INFO Plotting correl...
[ 2015-02-25 13:09:31 ] INFO Plotting pairwise...
[ 2015-02-25 13:09:31 ] INFO Plotting boxplot...
[ 2015-02-25 13:09:31 ] INFO Plotting boxplot...
[ 2015-02-25 13:09:31 ] INFO Plotting volcano...
[ 2015-02-25 13:09:31 ] INFO Contrast: WT_vs_TR
[ 2015-02-25 13:09:32 ] INFO Plotting biodist...
[ 2015-02-25 13:09:32 ] INFO Contrast: WT_vs_TR
[ 2015-02-25 13:09:32 ] INFO Plotting deheatmap...
[ 2015-02-25 13:09:32 ] INFO Contrast: WT_vs_TR
[ 2015-02-25 13:09:55 ] INFO Creating HTML report...
[ 2015-02-25 13:09:55 ] INFO Compressing figures...

Figures

Biotype detection

The biotype detection bar diagrams are a set of quality control charts that show the percentage of each biotype in the genome (i.e. in the whole set of features provided, for example, protein coding genes, non coding RNAs or pseudogenes) in grey bars, which proportion has been detected in a sample before normalization and after a basic filtering by removing features with zero counts in red lined bars, and the percentage of each biotype within the sample in solid red bars. The difference between grey bars and solid red bars is that the grey bars show the percentage of a feature in the genome while the solid red bars show the percentage in the sample. Thus, the solid red bars may be sometimes higher than the grey bars because certain features (e.g. protein coding genes) may be detected within a sample with a higher proportion relatively to their presence in the genome, as compared with other features. For example, while the percentage of protein coding genes in the whole genome is already higher than other biotypes, this percentage is expected to be even higher in an RNA-Seq experiment where one expects protein-coding genes to exhibit greater abundance. The vertical green line separates the most abundant biotypes (on the left-hand side, corresponding to the left axis scale) from the rest (on the right-hand side, corresponding to the right axis scale). Otherwise, the lower abundance biotypes would be indistinguishable. Unexpected outcomes in this quality control chart (e.g. very low detection of protein coding genes) would signify possible low quality of a sample.

Biotype detection counts

The biotype detection counts boxplots are a set of quality control charts that depict both the biological classification for the detected features and the actual distribution of the read counts for each biological type. The boxplot comprises a means of summarizing the read counts distribution of a sample in the form of a bar with extending lines, as commonly used way of graphically presenting groups of numerical data. A boxplot also indicates which observations, if any, might be considered outliers and is able to visually show different types of populations, without making any assumptions of the underlying statistical distribution. The spacing between the different parts of the box help indicate variance, skewness and identify outliers. The thick bar inside the colored box is the median of the observations while the box extends over the Interquartile Range of the observations. The whiskers extend up (down) to +/-1.5xIQR. Unexpected outcomes (e.g. protein coding read count distribution similar to pseudogene read count distribution) indicates poor sample quality.

Read and biotype saturation

The read and biotype saturation plots are a set of quality control charts that depict the read count saturation levels at several sequencing depths. Thus, they comprise a means of assessing whether the sequencing depth of an RNA-Seq experiment is sufficient in order to detect the biological features under investigation. These quality control charts are separated in two subgroups: the first subgroup (read saturation per biotype for all samples) is a set of plots, one for each biological feature (e.g. protein coding, pseudogene, lincRNA, etc.), that depict the number of detected features in different sequencing depths and for all samples in the same plot. The second subgroup (read saturation per sample for all biotypes) is a set of plots similar to the above, but with, there is one pair of plots with two panels for each sample, presenting all biological features. The left panel depicts the saturation levels for the less abundatnt features, while the right panel, the saturation for the more abundant features, as placing them all together would make the less abundant features indistinguishable. All the saturation plots should be interpreted as follows: if the read counts for a biotype tend to be saturated, the respective curve should tend to reach a plateau at higher depths. Otherwise, more sequencing is needed for the specific biotype.

Read saturation per biotype for all samples

Read saturation per sample for all biotypes

RNA-Seq reads noise

The read noise plots depict the percentage of biological features detected when subsampling the total number of reads. Very steep curves in read noise plots indicate that although the sequencing depth reaches its maximum, a relatively small percentage of total features is detected, indicating that the level of background noise is relatively high. Less steep RNA composition curves, indicate less noise. When a sample's curve deviate from the rest, it could indicate lower or higher quality, depending on the curves of the rest of the samples.

Correlation plots

The sample correlation plots depict the accordance among the RNA-Seq samples, as this is manifested through the read counts table used with the metaseqr pipeline, with two representations that both use the correlation matrix (a matrix which depicts all the pairwise correlations between each pair of samples) of the read counts matrix. The first is a correlation clustered heatmap which depicts the correlations among samples as color-scaled image and the hierarchical clustering tree depicts the grouping of the samples according to their correlation. Samples from the same group that are not clustered together provides an indication that there might be a quality problem with the dataset. The second is a 'correlogram' plot, where again the samples are hierarchically clustered and grouped but this time correlations are presented as ellipses inside each cell. Each cell represents a pairwise comparison and each correlation coefficient is represented by an ellipse whose 'diameter', direction and color depict the accordance for that pair of samples. Highly correlated samples are depicted as ellipses with narrow diameter, while poorly correlated samples are depicted as ellipses with wide diameters. Also, highly correlated samples are depicted as ellipses with a left-to-right upwards direction while poorly correlated samples are depicted as ellipses with a right-to-left upwards direction.

Correlation heatmap

Data correlogram

Pairwise scatterplots

The pairwise comparison plots are split in three parts: the upper diagonal consists of simple scatterplots for all pairwise sample comparisons, together with their Pearson correlation coefficient. It is a simple measure of between sample correlation using all the available data points instead of only the correlation matrix. The lower diagonal consists of mean-difference plots for all pairwise sample comparisons. A mean-difference plot (or a Bland-Altman plots) is a method of data plotting used in analyzing the agreement between two different assays/variables. In this graphical method the differences (or alternatively the ratios) between the two variables are plotted against the averages of the two. Such a plot is useful, for example, for analyzing data with strong correlation between x and y axes, when the (x,y) dots on the plot are close to the diagonal x=y. In this case, the value of the transformed variable X is about the same as x and y and the variable Y shows the difference between x and y. In both represantations, irregular shapes of the red smoother lines are an indication of poor correlation between samples or of other systematic bias sources, which is usually corrected through data normalization.

Boxplots

The boxplot comprises a means of summarizing the read counts distribution of a sample in the form of a bar with extending lines, as a commonly used way of graphically presenting groups of numerical data. A boxplot also indicates which observations, if any, might be considered outliers and is able to visually show different types of populations, without making any assumptions about the underlying statistical distribution. The spacings between the different parts of the box help indicate variance, skewness and identify outliers. The thick bar inside the colored box is the median of the observations while the box extends over the Interquartile Range of the observations. The whiskers extend up (down) to +/-1.5xIQR. Boxplots at similar levels indicate good quality of the normalization. If boxplots remain at different levels after normalization, maybe another normalization algorithm may have to be examined. The un-normalized boxplots show the need for data normalization in order for the data from different samples to follow the same underlying distribution and statistical testing becoming possible.

Boxplot of un-normalized data

Boxplot of normalized data

Volcano plots

A volcano plot is a scatterplot that is often used when analyzing high-throughput -omics data (e.g. microarray data, RNA-Seq data) to give an overview of interesting genes. The log2 fold change is plotted on the x-axis and the negative log10 p-value is plotted on the y-axis. A volcano plot combines the results of a statistical test (aka, p-values) with the magnitude of the change enabling quick visual identification of those genes that display large-magnitude changes that are also statistically significant. The horizontal dashed line sets the threshold for statistical significance, while the vertical dashed lines set the thresholds for biological significance. It should be noted that the volcano plots become harder to interpret when using more than one statistical algorithm and performing meta-analysis. This happens because the genes that have stronger evidence of being differentially expressed obtain lower p-values while the rest either remain at similar levels or obtain higher p-values. The result is a 'warped' volcano plot, with two main data clouds: one in the upper part of the plot, and one in the lower part of the plot. You can always zoom in when using interacting mode (the default).

Volcano plot for the contrast WT_vs_TR

DEG heatmaps

The Differentially Expressed Genes (DEGs) heatmaps depict how well samples from different conditions cluster together according to their expression values after normalization and statistical testing, for each requested statistical contrast. If samples from the same biological condition do not cluster together, this would constitute a warning sign regarding the quality of the samples. In addition, DEG heatmaps provide an initial view of possible clusters of co-expressed genes.

DEG heatmap for the contrast WT_vs_TR

Chromosome and biotype distributions of DEGs

The chromosome and biotype distributions bar diagram for Differentially Expressed Genes (DEGs) is split in two panels: i) on the left panel DEGs are distributed per chromosome and the percentage of each chromosome in the genome is presented in grey bars, the percentage of DEGs in each chromosome is presented in red lined bars and the percentage of certain chromosomes in the distribution of DEGs is presented in solid red bars. ii) on the right panel, DEGs are distributed per biotype and the percentage of each biotype in the genome (i.e. in the whole set of features provided, for example, protein coding genes, non coding RNAs or pseudogenes) is presented in grey bars, the percentage of DEGs in each biotype is presented in blue lined bars and the percentage of each biotype in DEGs is presented in solid blue lines. The vertical green line separates the most abundant biotypes (on the left-hand side, corresponding to the left axis scale), from the rest (on the right-hand side, corresponding to the right axis scale). Otherwise, the lower abundance, biotypes would be indistinguishable.

Biotype distribution in DEG for the contrast WT_vs_TR

Get all the figures in png, pdf format.

Results

DEG table for the contrast WT_vs_TR

The following table presents the top 10% statistically significant genes (use the download links below the table to retrieve the whole list) for the contrast WT_vs_TR. The fields of the table correspond to the requested features to be exported. The table can be searched using the search field on the top right.

chromosome	start	end	gene_id	gc_content	strand	gene_name	biotype	p-value_edger	FDR_edger	log2_normalized_fold_change_WT_vs_TR	log2_normalized_mean_counts_WT	log2_normalized_median_counts_WT	log2_normalized_sd_counts_WT	log2_normalized_cv_counts_WT	log2_normalized_mean_counts_TR	log2_normalized_median_counts_TR	log2_normalized_sd_counts_TR	log2_normalized_cv_counts_TR	log2_normalized_counts_SCR2	log2_normalized_counts_SCR1
chr8	112514335	112523704	NM_146214	0.4535	+	Tat	protein_coding	1.13e-31	1.14e-27	-8.447083	9.45	9.45	NA	NA	1.58	1.58	NA	NA	9.45	1.58
chr13	62966112	62989642	NM_019395	0.4759	-	Fbp1	protein_coding	1.41e-22	7.14e-19	-5.290940	9.61	9.61	NA	NA	4.39	4.39	NA	NA	9.61	4.39
chr18	21230844	21258700	NM_008586	0.4271	+	Mep1b	protein_coding	2.20e-22	7.44e-19	-5.222392	9.81	9.81	NA	NA	4.64	4.64	NA	NA	9.81	4.64
chr15	82620536	82624675	NM_029562	0.5196	-	Cyp2d26	protein_coding	1.64e-21	4.15e-18	-4.924973	10.14	10.14	NA	NA	5.25	5.25	NA	NA	10.14	5.25
chr9	78178826	78194952	NM_008182	0.4604	-	Gsta2	protein_coding	1.43e-19	2.89e-16	-4.645604	9.69	9.69	NA	NA	5.09	5.09	NA	NA	9.69	5.09
chr1	90303593	90323013	NM_029269	0.4581	+	Spp2	protein_coding	2.60e-19	4.38e-16	-4.599511	9.81	9.81	NA	NA	5.25	5.25	NA	NA	9.81	5.25
chr19	8692485	8702789	NM_008766	0.4858	+	Slc22a6	protein_coding	3.95e-18	5.71e-15	-4.245625	10.29	10.29	NA	NA	6.07	6.07	NA	NA	10.29	6.07
chr10	88348736	88392260	NM_145423	0.3962	+	Slc5a8	protein_coding	4.69e-16	5.94e-13	-4.233251	8.94	8.94	NA	NA	4.75	4.75	NA	NA	8.94	4.75
chr9	44607482	44611034	NM_138951	0.5283	-	Ttc36	protein_coding	3.38e-15	3.80e-12	-3.977280	9.15	9.15	NA	NA	5.21	5.21	NA	NA	9.15	5.21
chr19	6535855	6543070	NM_009203	0.5572	-	Slc22a12	protein_coding	4.42e-14	4.47e-11	-3.592955	9.82	9.82	NA	NA	6.25	6.25	NA	NA	9.82	6.25
chr3	98356473	98367961	NM_008294	0.4190	-	Hsd3b4	protein_coding	9.48e-14	8.73e-11	-3.441284	10.76	10.76	NA	NA	7.33	7.33	NA	NA	10.76	7.33
chr13	55501008	55516056	NM_011392	0.5065	+	Slc34a1	protein_coding	1.11e-13	9.37e-11	-3.361163	12.55	12.55	NA	NA	9.19	9.19	NA	NA	12.55	9.19
chr2	172978573	172984754	NM_011044	0.5163	+	Pck1	protein_coding	1.50e-13	1.17e-10	-3.526368	9.59	9.59	NA	NA	6.09	6.09	NA	NA	9.59	6.09
chr9	45214159	45218361	NM_007503	0.5353	+	Fxyd2	protein_coding	4.86e-13	3.52e-10	-3.268351	11.89	11.89	NA	NA	8.63	8.63	NA	NA	11.89	8.63
chr7	31278676	31298421	NM_019546	0.4983	+	Prodh2	protein_coding	4.17e-12	2.81e-09	-3.327165	9.11	9.11	NA	NA	5.81	5.81	NA	NA	9.11	5.81
chr4	49548866	49562355	NM_144903	0.4256	-	Aldob	protein_coding	6.21e-12	3.93e-09	-3.069814	12.55	12.55	NA	NA	9.49	9.49	NA	NA	12.55	9.49
chr8	62984920	63024474	NM_011834	0.3988	+	Aadat	protein_coding	7.35e-12	4.38e-09	-3.184572	9.80	9.80	NA	NA	6.63	6.63	NA	NA	9.80	6.63
chr13	99404932	99407365	NM_026685	0.5004	-	Tmem174	protein_coding	4.45e-11	2.50e-08	-3.055754	9.58	9.58	NA	NA	6.54	6.54	NA	NA	9.58	6.54
chr11	75245237	75253003	NM_008878	0.5186	-	Serpinf2	protein_coding	2.10e-10	1.12e-07	-2.895185	9.94	9.94	NA	NA	7.06	7.06	NA	NA	9.94	7.06
chr12	112509321	112514637	NM_033603	0.6028	+	Amn	protein_coding	2.85e-10	1.44e-07	-2.906891	9.52	9.52	NA	NA	6.63	6.63	NA	NA	9.52	6.63
chr14	56372891	56377261	NM_009894	0.4907	-	Cideb	protein_coding	5.93e-10	2.86e-07	-2.866057	9.29	9.29	NA	NA	6.44	6.44	NA	NA	9.29	6.44

Download the DEG result list for WT_vs_TR.
Download the whole result list for WT_vs_TR.

Download the raw read counts table for the experiment.
Download the normalized read counts table for the experiment.

References

Statham, A.L., Strbenac, D., Coolen, M.W., Stirzaker, C., Clark, S.J., Robinson, M.D. (2010). Repitools: an R package for the analysis of enrichment-based epigenomic data. Bioinformatics 26(13), 1662-1663.
Robinson, M.D., McCarthy, D.J., and Smyth, G.K. (2010). edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139-140.
Planet, E., Attolini, C.S., Reina, O., Flores, O., and Rossell, D. (2012). htSeqTools: high-throughput sequencing quality control, processing and visualization in R. Bioinformatics 28, 589-590.
Tarazona, S., Garcia-Alcalde, F., Dopazo, J., Ferrer, A., and Conesa, A. (2011). Differential expression in RNA-seq: a matter of depth. Genome Res 21, 2213-2223.
Benjamini, Y., and Hochberg, Y. (1995). Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B (Methodological) 57, 289-300.