Analysis types

BAM QC

BAM QC reports information for the evaluation of the quality of the provided alignment data (a BAM file). In short, the basic statistics of the alignment (number of reads, coverage, GC-content, etc.) are summarized and a number of useful graphs are produced. This analysis can be performed with any kind of sequencing data, e.g. whole-genome sequencing, exome sequencing, RNA-seq, ChIP-seq, etc.

In addition, it is possible to provide an annotation file so the results are computed for the reads mapping inside (and optionally outside) of the corresponding genomic regions, which can be especially useful for evaluating target-enrichment sequencing studies.

To start a new BAM QC analysis activate main menu item File ‣ New Analysis ‣ BAM QC.

Examples

Input Parameters

BAM file
Path to the sequence alignment file in BAM format. Note, that the BAM file has to be sorted by chromosomal coordinates. Sorting can be performed with samtools sort.
Analyze regions
Activating this option allows the analysis of the alignment data for the regions of interest.
Regions file(GFF/BED file)
The path to the annotation file that defines the regions of interest. The file must be tab-separated and have GFF/GTF or BED format.

Note

A typical problem when working with human genome annotations is the inconsistency between chromosome names due to “chr” prefix. For example, Ensemble annotations do not include this prefix, while UCSC annotations do. This can become a problem when asscociating regions file with the BAM alignment. Qualimap handles this problem: if the reference sequence of a region has “chr” prefix, it tries to search for sequence name with prefix and without prefix.

Library strand specificity

The sequencing protocol strand specificity: non-strand-specific, forward-stranded or reverse-stranded. This information is required to calculate the number of correct strand reads.
Analyze outside regions
If checked, the information about the reads that are mapped outside of the regions of interest will be also computed and shown in a separate section.
Chromosome limits
If selected, vertical dotted lines will be placed at the beginning of each chromosome according to the information found in the header of the BAM file.
Compare GC content distribution with
This allows to compare the GC distribution of the sample with the selected pre-calculated genome GC distribution. Currently two genome distributions are available: human (hg19,hg38) and mouse (mm9,mm10). More species will be included in future releases.
Detect overlapping paired-end reads
In case of small insert size the paired-end read alignmetns might overlap in high proportion. Using this option detection of overlapping pairs can be activated. Additionally, adapted mean coverage is calcualted based on extraction of pair overlap-region.
Skip duplicates
This option allows to skip duplicate alignments from analysis. There are three modes of this option. By default, the duplicates are skipped only if they are flagged in BAM file and remaining alignments are futher analyzed by Qualimap. Additionally it is possible to skip only the duplicates detected by Qualimap method (based on duplication rate estimation) or aplly both approaches. Number of skipped duplicates will be shown in the report.
Targeted coverage hist. cut point(X)
This optinon allows to extend per bin coverage histogram cut point. Default is 50X.
Duplication hist. cut point(X)
This optinon allows to extend duplication rate histogram maximum value. Default is 50.

Advanced parameters

Number of windows
Number of windows used to split the reference genome. This value is used for computing the graphs that plot information across the reference. Basically, reads falling in the same window are aggregated in the same bin. The higher the number, the bigger the resolution of the plots but also longer time will be used to process the data. By default 400 windows are used.
Homopolymer size
Only homopolymers of this size or larger will be considered when estimating homopolymer indels count.
Number of threads
In order to speed up the computation, the BAM QC analysis computation can be performed in parallel on a multicore system using the given number of threads. More information on the parallelization of qualimap can be found in FAQ. The default number of threads equals number of available processors.
Size of the chunk
In order to reduce the load of I/O, reads are analyzed in chunks. Each chunk contains the selected number of reads which will be loaded into memory and analyzed by a single thread. Smaller numbers may result in lower performance, but also the memory consumption will be reduced. The default value is 1000 reads.

Output

Summary

Basic information and statistics for the alignment data. The following sections are available:

Globals

This section contains information about the total number of reads, number of mapped reads, paired-end mapping performance, read length distribution, number of clipped reads and duplication rate (estimated from the start positions of read alignments).

ACGT Content

Nucleotide content and GC percentage in the mapped reads.

Coverage

Mean and standard deviation of the coverage depth.

Mapping quality

Mean mapping quality of the mapped reads.

Insert size

Mean, standard deviation and percentiles of the insert size distribution if applicable. The features are computed based on the TLEN field of the SAM file.

Mismatches and indels

The section reports general alignment error rate (computed as a ratio of total collected edit distance to the number of mapped bases), total number of mismatches and total number of indels (computed from the CIGAR values). Additionally fraction of the homopolymer indels among total indels is provided. Note, the error rate and mismatches metrics are based on optional fields of a SAM record (NM for edit distance, MD for mismatches). The features are not reported if these fields are missing in the SAM file.

Chromosome stats

Number of mapped bases, mean and standard deviation of the coverage depth for each chromosome as defined by the header of the SAM file.

For region-based analysis the information is given inside of regions, including some additional information like, for example, number of correct strand reads.

Input

Here one can check the input data and the parameters used for the analysis.

Coverage Across Reference

This plot consists of two figures. The upper figure provides the coverage distribution (red line) and coverage deviation across the reference sequence. The coverage is measured in X [1]. The lower figure shows GC content across reference (black line) together with its average value (red dotted line).

Coverage Histogram

Histogram of the number of genomic locations having a given coverage rate. The bins of the x-axis are conveniently scaled by aggregating some coverage values in order to produce a representative histogram also in presence of the usual NGS peaks of coverage.

Coverage Histogram (0-50X)

Histogram of the number of genomic locations having a given coverage rate. In this graph genome locations with a coverage greater than 50X are grouped into the last bin. By doing so a higher resolution of the most common values for the coverage rate is obtained.

Genome Fraction Coverage

Provides a visual way of knowing how much reference has been sequenced with at least a given coverage rate. This graph should be interpreted as in this example:

If one aims a coverage rate of at least 25X (x-axis), how much of reference (y-axis) will be considered? The answer to this question in the case of the whole-genome sequencing provided example is ~83%.

Duplication Rate Histogram

This plot shows the distribution of duplicated read starts. Due to several factors (e.g. amount of starting material, sample preparation, etc) it is possible that the same fragments are sequenced several times. For some experiments where enrichment is used (e.g. ChIP-seq ) this is expected at some low rate. If most of the reads share the exact same genomic positions there is very likely an associated bias.

Mapped Reads Nucleotide Content

This plot shows the nucleotide content per position of the mapped reads.

Mapped Reads GC Content Distribution

This graph shows the distribution of GC content per mapped read. If compared with a precomputed genome distribution, this plot allows to check if there is a shift in the GC content.

Mapped Reads Clipping Profile

Represents the percentage of clipped bases across the reads. The clipping is detected via SAM format CIGAR codes ‘H’ (hard clipping) and ‘S’ (soft clipping). In addition, the total number of clipped reads can be found in the report Summary. The plot is not shown if there are no clipped-reads are found. Total number of clipped reads can be found in Summary. Example.

Homopolymer Indels

This bar plot shows separately the number of indels that are within a homopolymer of A’s, C’s, G’s or T’s together with the number of indels that are not within a homopolymer. Large numbers of homopolymer indels may indicate a problem in a sequencing process. An indel is considered homopolymeric if it is found within a homopolymer (defined as at least 5 equal consecutive bases). Owing to the fact that Qualimap works directly from BAM files (and not from reference genomes), we make use of the CIGAR code from the corresponding read for this task. Indel statistics cam be found in a dedicated section of the Summary report.

This chart is not shown if the sample doesn’t contain any indels.

Mapping Quality Across Reference

This plot provides the mapping quality distribution across the reference. To construct the plot mean mapping quality is computed for each window.

Mapping Quality Histogram

Histogram of the number of genomic locations having a given mapping quality. To construct the histogram mean mapping quality is computed at each genome position with non-zero coverage and collected. According to Specification of the SAM format the range for the mapping quality is [0-255].

Insert Size Across Reference

This plot provides the insert size distribution across the reference. Insert size is collected from the SAM alignment field TLEN. Only positive values are taken into account. To construct the plot mean insert size is computed for each window.

Insert Size Histogram

Histogram of insert size distribution. To construct the histogram all collected insert size values (number of read alignments of a certain insert size) are seprated in bins. The default number of bins is 50. The number of reads in the bin will be the sum from insert size of the bins. The detailed values for insert are reported in raw data report.

RNA-seq QC

RNA-seq QC reports quality control metrics and bias estimations which are specific for whole transcriptome sequencing, including reads genomic origin, junction analysis, transcript coverage and 5’-3’ bias computation. This analysis could be applied as a complementary tool together with BAM QC and additionally to produce gene counts for further analysis with Counts QC.

To start a new RNA-seq QC analysis activate main menu item File ‣ New Analysis ‣ RNA-seq QC.

Examples

  • RNA-seq QC report. This report was produced using the RNA-seq alignment of Homo sapiens kidney sample [Marioni] and Ensembl v.64 GTF file.
  • These data can be downloaded from here.

Input parameters

BAM file
Path to the sequence alignment file in BAM format, produced by a splicing-aware aligner similar to Tophat.
GTF file
Genomic annotations in Ensembl GTF format. The corresponding annotations can be downloaded from the Ensembl website.

Note

Only annotations in GTF format are supported for this analysis mode. GTF annotations allow to reconstruct the exon structure of transcripts to compute the coverage. For simple region-based analysis please use BAM QC.

Library protocol
The strand-specficity of the sequencing library. By default non-strand specific library is assumed.
Paired-end analysis
This option activates counting of pair fragments instead of counting of single reads. Only valid for paired-end sequencing experiments.
Alignment sorted by name
The paired-end analysis requires the BAM file to be sorted by name. If the BAM file is already sorted by name, then this option should be check, otherwise temporary BAM sorted by name will be created.
Output counts
If checked, the gene counts will be saved to a specified file.
Path to counts
Path to the output file with the computed counts.

Advanced parameters

Multi-mapped reads
Select method to count reads that are mapped to several genome locations. By default only uniquely-mapped-reads are used to compute counts. However, it is possible to include multimapped reads by activating proprtional method. More details here.

Output

Summary

The summary contains the following sections:

Reads alignment

The assignment of read counts per-category:
  • total number of mapped reads (left/right in case of paired-end reads, secondary alignments are ignored)
  • total number of alignments
  • number of secondary alignments ( SAM flag for multi-mapped reads )
  • number of non-unique alignments ( SAM format “NH” tag of a read is more than one, by default not taken into account during further analysis )
  • number of reads aligned to genes
  • number of ambiguous alignments (belong to several genes, ignored during counting procedure)
  • number of alignments without any feature (intronic and intergenic)
  • number of ignored alignments when the chromsome is not found in annotation
  • number of unmapped reads.

Additionally, if default non-strand-specific protocol is stated in the settings, forward- and reverse-strand-specificitiy is estimated based on the strands of the read alginments. Similarity of computed proportions (i.e ~0.5 forward, ~0.5 reverse) confirms non-strand-specificity, while domination of a certain strand estimates possible protocol ( i.e. reverse SSP in case of ~0.1 forward, ~0.9 reverse)

Transcript coverage profile

The profile provides ratios between mean coverage at the 5’ region, the 3’ region and the whole transcript. The 5’ bias is the ratio between mean coverage at the 5’ region and the whole transcript, while the 3’ bias is the ratio between mean coverage at the 3’ region and the whole transcript. 5’-3’ bias is the ratio between both biases.

To compute these values for each transcript mean coverage along with mean coverage in first 100 bp (5’ region) and last 100 bp (3’region) are calculated and collected. Afterwards, the collected values are sorted and median is selected from each array to compute the ratios.

Reads genomic origin

Shows how many alignments fall into exonic, intronic and intergenic regions along with a number of intronic/intergenic alignments overlapping exons. Exonic region includes 5’UTR,protein coding region and 3’UTR region.

Junction analysis

Total number of reads with splice junctions and 10 most frequent junction rates.

Input

Here one can check the input data and the parameters used for the analysis.

Reads Genomic Origin

Pie chart showing how many of read alignments fall into exonic, intronic and intergenic regions.

Coverage Profile (Total)

The plot shows mean coverage profile of the transcripts. All transcripts with non-zero coverage are used to calculate this plot.

Coverage Profile (Low)

The plot shows mean coverage profile of 500 lowest-expressed genes.

Coverage Profile (Total)

The plot shows mean coverage profile of 500 highest-expressed genes.

Coverage Histogram (0-50x)

Coverage of transcripts from 0 to 50X. If certain genes have higher coverage level they are added to the last column (50X).

Junction Analysis

This pie chart shows analysis of junction positions in spliced alignments. Known category represents percentage of alignments where both junction sides are known. Partly known represents alignments where only one junction side is known. All other alignments with junctions are marked as Novel.

Counts QC

In RNA-seq experiments, the reads are usually first mapped to a reference genome. It is assumed that if the number of reads mapping to a certain biological feature of interest (gene, transcript, exon, ...) is sufficient, it can be used as an estimation of the abundance of that feature in the sample and interpreted as the quantification of the expression level of the corresponding region.

These count data can be utilized for example to assess differential expression between two or more experimental conditions. Before assessing differential expression analysis, researchers should be aware of some potential limitations of RNA-seq data, as for example: Has the saturation been reached or more features could be detected by increasing the sequencing depth? Which type of features are being detected in the experiment? How good is the quantification of expression in the sample? All of these questions are answered by interpreting the plots generated by Counts QC.

Starting from version 2.0 Counts QC module has been redisigned to work with multiple samples under different conditions. The new functionality is based on NOISeq package, therefore to use Counts QC it is required to have R language along with NOISeq and optparse packages installed.

To run this analysis activate from the main menu File ‣ New Analysis ‣ Counts QC.

Note

If count data need to be generated, one can use the provided tool Compute counts.

Example

  • RNA-seq counts analysis from 2 experiments can be found here
  • Sample counts data can be downloaded from here.

Input Parameters

Samples

The input samples can be added using button Add.

For each input sample it is required to provide the following information:

  • Sample name. Name of the analyzed sample as it will be used as a legend in the plots.
  • Path to the input file containing the counts data for the sample. This must be a tab-delimited file with at least two columns. First column of the file must contain feature IDs, while other columns should contain counts for features. Rows starting with # symbol and empty lines are ignored.
  • Data column index. By default it is assumed that the counts are contained in the second column of the input file. However if the input file contains counts for multiple samples it is possible to define the column corresponding for the sample.
  • Condition index. If comparison of conditions is activated, this index defines under which condition was the input sample.

Each added sample will be shown in Samples table. One can edit samples using button Edit and remove them using button Remove.

Counts threshold

In order to remove the influence of spurious reads, a feature is considered as detected if its corresponding number of counts is greater than this threshold. By default, the threshold value is set to 5 counts, meaning that features having less than 5 counts will not be taken into account.

Compare conditions

This option allows to compare groups of samples under different conditions. The name of a specific condition can be given using field Condition name.

Note

Currently Qualimap allows to compare samples under two conditions. More conditions will be supported in future versions.

Include feature classification

Optional. This option enables analysis of distribution of counts among feature groups defined by the biotype. In addition GC-content and length bias will be estimated.

Species

For convinience, Qualimap provides the Ensembl annotations for certain species (currently Human and Mouse). In order to use these annotations, Ensembl Gene IDs should be used as the feature IDs on the count files (e.g. ENSG00000251282). If this is true, mark the box to enable this option and select the corresponding species. More annotations and species will be made available in future releases.

Info File

File containing annotations of the features of the count files. It must be a four-column tab-delimited text file, with the features names or IDs in the first column, the group (e.g. the biotype from Ensembl database) in the second column, feature length in the third column and feature GC-content in the last column (see human.ens68.txt for an example). Please, make sure that the features IDs on this file are the same in the count files.

Note

To generate info file based on an arbitrary GTF annotations and genome FASTA file, one can use the following Python script available from Qualimap repo.

Output

Many of plots in Counts QC mode are created using NOISeq package. The NOISeq vignette contains a lot of useful information about the plots and how to interpret them. Here we provide short explanation of the plots.

Global Plots

Plots from this report present a global overview of the counts data and include all samples.

Counts Density

This plot shows density of counts computed from the histogram of log-transformed counts. In order to avoid infinite values in case of zero counts the transformation log2(expr + 0.5) is applied, where expr is a number of read counts for a given feature. Only log-transformed counts having value greater than 1 are plotted.

Scatterplot Matrix

The panel shows a scatterplot along with smoothed line (lower panel) and Pearson correlation coefficients (upper panel) for each pair of samples. Plots are generated using log-transformed counts.

Saturation

This plot provides information about the level of saturation in the samples, so it helps the user to decide if more sequencing is needed and more features could be detected when increasing the number of reads. These are some tips for the interpretation of the plot:

  • The increasing sequencing depth of the sample is represented at the x-axis. The maximum value is the real sequencing depth of the sample(s). Smaller sequencing depths correspond to samples randomly generated from the original sample(s).
  • The curves are associated to the left y-axis. They represent the number of detected features at each of the sequencing depths in the x-axis. By “detected features” we refer to features with more than k counts, where k is the Count threshold selected by the user.
  • The bars are associated to the right y-axis. They represent the number of newly detected features when increasing the sequencing depth in one million reads at each sequencing depth value.

Counts Distribution

This box plot shows the global distribution of counts in each sample.

Features With Low Counts

This plot shows the proportion of features with low counts in the samples. Such features are usually less reliable and could be filtered out. In this plot, the bars show the percentage of features within each sample having more than 0 counts per million (CPM), or more than 1, 2, 5 and 10 CPM.

Individual Sample Plots

Apart from global overview there are plots generated individually for each sample.

Saturation

For each sample, a saturation plot is generated like the one described in Global Saturation.

When a Info File is provided by the user or annotations are chosen from those supplied by Qualimap, additional series of plots are generated:

Bio Detection

This barplot allows the user to know which kind of features are being detected his sample(s). The x-axis shows all the groups included in the annotations file. The gray bars are the percentage of features of each group within the reference genome (or transcriptome, etc.). The striped color bars are the percentages of features of each group detected in the sample with regard to the genome. The solid color bars are the percentages that each group represents in the total detected features in the sample.

Counts Per Biotype

A boxplot per each group describes the counts distribution in the given biotype.

Length Bias

The plot describes the relationship between the length of the features and the expression values. The length is divided into bins. Mean expression of features falling into a particular length interval is computed and plotted. A cubic spline regression model is fitted to explain the relation between length and expression. Coefficient of determination R^2 and p-value are shown together with regression curve.

GC Bias

The plot describes the relantionship between the GC-content of the features and the expression values. The data for the plot is generated similar to Length Bias plot. The GC content divided into beans and then mean expression of features corresponding to given GC interval are computed. The relation between GC-content and expression is investigated using cubic spline regression model.

Comparison Plots

When Compare conditions option is selected, additional plots comparing data in groups of samples having the same biological condition or treatment are available.

Counts Distribution

The plot is similar to the one in Global report. It compares distributions of mean counts across conditions.

Features With Low Counts

The plot is similar to the one in Global report. It compares proportions of features with low counts using mean counts across conditions.

Bio Detection

The plot is similar to the one in Indvidual Sample Plots report. It compares distribution of the detected features for the given biotype for mean counts across conditions.

Length Bias

The plot is similar to the one in Individual Sample Plots report. It analyzes relation between feature length and expression across conditions.

GC Bias

The plot is similar to the one in Individual Sample Plots report. It analyzes realtion between GC-content and expression across conditions.

Multi-sample BAM QC

Very often in genomics one has to work with multiple samples, which could represent sequencing results from either biological replicates or different conditions. For example, to reliably detect significant mutations from sequencing data in cancer it is required to analyze tens or even hundreds of samples from matched normal-tumor data. When performing such large scale experiments it is always important to know if all samples belonging to a specific group pass the quality controls. To detect possible outliers one can compare results of BAM QC analysis performed on each individual sample.

QualiMap provides an automated solution for this task. Basically, the QC metrics computed in BAM QC analysis are combined together for all samples. Additionally Principal Component Analysis is performed to analyze variability and detect outliers.

Note

Starting from version 2.2 it is possible to assign groups marking biolobical or technical conditions of the samples.

One can apply multi-sample analysis for precomputed results of QualiMap BAM QC or directly for raw BAM files. In latter case firstly BAM QC analysis will be performed for each input file and then multi-sample analysis will be executed.

To start a new multi-sample BAM QC analysis activate main menu item File ‣ New Analysis ‣ Multisample BAM QC.

Examples

See the Sample data section for more details about the data used in the example.

Input Parameters

There are 2 types of input data that are accepted by Multi-sample BAM QC:

  1. By default directory with the summary statistics and plot data produced by BAM QC analysis is expected as input data for multi-sample comparison.
  2. If a special “raw data” mode is activated, then BAM files can be provided as input. In this case Qualimap will first run the BAM QC analysis on each indvidual BAM file, and then multi-sample report will be computed.

The input samples can be added using button Add. For each sample one has to provide the following information:

  1. Name of the sample as it will be used in legend.
  2. Path to the folder with which contains results of BAM QC analysis performed on the sample. The folder must include file genome_results.txt and subfolder raw_data_qualimapReport containing data of BAM QC plots. If “Raw data” mode is activated then the path to the BAM file should be provided.
  3. Group of the sample. This option allows to combine the samples of the same condition. After the group is assigned, the samples in the plots belonging to the group will have the same color. Importantly, if the groups are avaialble, they should be provided for each sample. Empty value will mean no group.

Note

In QualiMap version <= 2.0 directory with raw data of BAM QC analysis was called raw_data. This name is also supported.

Each added sample will be shown in Samples table. One can edit samples using button Edit and remove them using button Remove.

Additionally it is possible to import configuration file, that is applied for command line interface using button Import configuration.... The configuration file is explained in the overview of the command line mode.

“Raw data” mode: run BAM QC on input samples

Activate this checkbox to analyze BAM files directly. A selected set of options is available to customize BAM QC process. One can read detailed explantion of these options in a corresponding section of the manual.

To start the analysis click button Run analysis.

Output

Summary

The summary table contains comparison of selected critical alignment metrics for all samples. The metrics include mean and standard deviation of coverage, mean GC content, mean insert size and mean mapping qualities. If the sample groups are provided, they are also shown for each sample.

Input

Here one can check the input data and the parameters used for the analysis.

PCA

The alignment features presented in the Summary section undergo Principal Component Analysis. Afterwards the biplot presenting first and second principal component is constructed. The plot shows how much variability demonstarte the analyzed samples. It allows to detect if any samples group together and if there are any outliers among analyzed samples.

Coverage Across Reference, Coverage Histogram (0-50X) , Genome Fraction Coverage, Duplication Rate Histogram, Mapped Reads GC Content, Mapped Reads GC Content Distribution, Mapped Reads Clipping Profile, Mapping Quality Across Reference, Mapping Quality Histogram, Insert Size Across Reference, Insert Size Histogram

The following plots demonstrate the comparison of samples using data from corresponding plots computed during BAM QC analysis. Each curve on a plot represents a single sample.

Please refer to documentation of BAM QC for detailed information about the plots.

***

[1]Example for the meaning of X: If one genomic region has a coverage of 10X, it means that, on average, 10 different reads are mapped to each nucleotide of the region.
[2]Downloaded from Biomart v.61.
[Marioni]Marioni JC et al, “RNA-seq: An assessment of technical reproducibility and comparison with gene expression arrays”. Genome Res. 2008. 18: 1509-1517.