* Merging and mapping of overlapping paired-end reads with new options --peOverlapNbasesMin and --peOverlapMMp. The developmment of this algorithm was supported by Illumina, Inc. Many thanks to June Snedecor, Xiao Chen, and Felix Schlesinger for their extensive help in developing this feature.
* --varVCFfile option to input variant VCF file.
* New SAM attributes in the --outSAMattributes, vG, vA, and vW to report variants overlapping alignments.
* --waspOutputMode option for filtering allele specific alignments. This is re-implementation of the original WASP algorithm by Bryce van de Geijn, Graham McVicker, Yoav Gilad & Jonathan K Pritchard. Please cite the original WASP paper: Nature Methods 12, 1061–1063 (2015), https://www.nature.com/articles/nmeth.3582 . Many thanks to Bryce van de Geijn for fruitful discussions.
* Detection of multimapping chimeras, with new options --chimMultimapNmax, --chimMultimapScoreRange and --chimNonchimScoreDropMin . Many thanks to Brian Haas for testing and feedback.
Minor new features:
-------------------
* --alignInsertionFlush option which defines how to flush ambiguous insertion positions: None: old method, insertions are not flushed; Right: insertions are flushed to the right.
* --outSAMtlen option to select the calculation method for the TLEN field in the SAM/BAM files.
* --outBAMsortingBinsN option to control the number of sorting bins. Increasing this number reduces the amount of RAM required for sorting.
**1. Merging and mapping of overlapping paired-end reads.**
This feature improves mapping accuracy for paired-end libraries with short insert sizes, where many reads have overlapping mates. Importantly, it allows detection of chimeric junction in the overlap region.
STAR will search for an overlap between mates larger or equal to --peOverlapNbasesMin bases with proportion of mismatches in the overlap area not exceeding --peOverlapMMp .
If the overlap is found, STAR will map merge the mates and attempt to map the resulting (single-end) sequence.
If requested, the chimeric detection will be performed on the merged-mate sequence, thus allowing chimeric detection in the overlap region.
If the score of this alignment higher than the original one, or if a chimeric alignment is found, STAR will report the merged-mate aligment instead of the original one.
In the output, the merged-mate aligment will be converted back to paired-end format.
The developmment of this algorithm was supported by Illumina, Inc.
Many thanks to June Snedecor, Xiao Chen, and Felix Schlesinger for their extensive help in developing this feature.
**2. Detection of personal variants overlapping alignments.**
Option --varVCFfile /path/to/vcf/file is used to input VCF file with personal variants. Only single nucleotide variants (SNVs) are supported at the moment.
Each variant is expected to have a genotype with two alleles.
To output variants that overlap alignments, vG and vA have to be added to --outSAMattributes list.
SAM attribute vG outputs the genomic coordinate of the variant, allowing for identification of the variant.
SAM attribute vA outputs which allele is detected in the read: 1 or 2 match one of the genotype alleles, 3 - no match to genotype.
**3. WASP filtering of allele specific alignments.**
This is re-implementation of the original WASP algorithm by Bryce van de Geijn, Graham McVicker, Yoav Gilad & Jonathan K Pritchard. Please cite the original [WASP paper: Nature Methods 12, 1061–1063 (2015) ](https://www.nature.com/articles/nmeth.3582).
WASP filtering is activated with --waspOutputMode SAMtag, which will add vW tag to the SAM output:
vW:i:1 means alignment passed WASP filtering, while all other values mean it did not pass.
Many thanks to Bryce van de Geijn for fruitful discussions.
**4. Detection of multimapping chimeras.**
Previous STAR chimeric detection algorithm only detected uniquely mapping chimeras, which reduced its sensitivity in some cases.
The new algorithm can detect and output multimapping chimeras. Presently, the only output into Chimeric.out.junction is supported.
This algorithm is activated with >0 value in --chimMultimapNmax, which defines the maximum number of chimeric multi-alignments.
The --chimMultimapScoreRange (=1 by default) parameter defines the score range for multi-mapping chimeras below the best chimeric score, similar to the --outFilterMultimapScoreRange parameter for normal alignments.
The --chimNonchimScoreDropMin (=20 by default) defines the threshold triggering chimeric detection: the drop in the best non-chimeric alignment score with respect to the read length has to be smaller than this value.
Many thanks to Brian Haas for testing and feedback.
Minor new features:
-------------------
* --outSAMtlen 1/2 option to select the calculation method for the TLEN field in the SAM/BAM files:
1 ... leftmost base of the (+)strand mate to rightmost base of the (-)mate. (+)sign for the (+)strand mate
2 ... leftmost base of any mate to rightmost base of any mate. (+)sign for the mate with the leftmost base. This is different from 1 for overlapping mates with protruding ends
* --alignInsertionFlush option which defines how to flush ambiguous insertion positions: None: old method, insertions are not flushed; Right: insertions are flushed to the right.
* --outBAMsortingBinsN option to control the number of sorting bins. Increasing this number reduces the amount of RAM required for sorting.
@@ -72,7 +72,7 @@ STAR is compiled with gcc c++ compiler and depends only on standard gcc librarie
}
\paragraph{Mac OS X.\newline}
Current versions of Mac OS X Xcode are shipped with Clang replacing the standard gcc compiler. Presently, standard Clang does not support OpenMP which creates problems for STAR compilation. One option to avoid this problem is to install gcc (preferrably using \code{homebrew} package manager). Another option is to add OpenMP functionality to Clang.
Current versions of Mac OS X Xcode are shipped with Clang replacing the standard gcc compiler. Presently, standard Clang does not support OpenMP which creates problems for STAR compilation. One option to avoid this problem is to install gcc (preferably using \code{homebrew} package manager). Another option is to add OpenMP functionality to Clang.
\subsection{Basic workflow.}
Basic STAR workflow consists of 2 steps:
@@ -153,7 +153,7 @@ Note, that the \opt{sjdbFileChrStartEnd} file can contain duplicate (identical)
For small genomes, the parameter \opt{genomeSAindexNbases}\textbf{must} to be scaled down, with a typical value of \code{min(14, log2(GenomeLength)/2 - 1)}. For example, for 1~megaBase genome, this is equal to 9, for 100~kiloBase genome, this is equal to 7.
\subsubsection{Genome with a large number of references.}
If you are using a genome with a large (\textgreater 5,000) number of references (chrosomes/scaffolds), you may need to reduce the \opt{genomeChrBinNbits} to reduce RAM consumption. The following scaling is recommended: \opt{genomeChrBinNbits} = \code{min(18,log2(GenomeLength/NumberOfReferences))}. For example, for 3~gigaBase genome with 100,000 chromosomes/scaffolds, this is equal to 15.
If you are using a genome with a large (\textgreater 5,000) number of references (chrosomes/scaffolds), you may need to reduce the \opt{genomeChrBinNbits} to reduce RAM consumption. The following scaling is recommended: \opt{genomeChrBinNbits} = \code{min(18,log2[max(GenomeLength/NumberOfReferences,ReadLength)])}. For example, for 3~gigaBase genome with 100,000 chromosomes/scaffolds, this is equal to 15.
@@ -449,6 +449,48 @@ This is the original 2-pass method which involves genome re-generation step in-b
\item Run the 2nd pass mapping for all samples with the new genome index.
\end{enumerate}
\section{Merging and mapping of overlapping paired-end reads.}
This feature improves mapping accuracy for paired-end libraries with short insert sizes, where many reads have overlapping mates. Importantly, it allows detection of chimeric junction in the overlap region.
STAR will search for an overlap between mates larger or equal to \opt{peOverlapNbasesMin} bases with proportion of mismatches in the overlap area not exceeding \opt{peOverlapMMp}.
If the overlap is found, STAR will map merge the mates and attempt to map the resulting (single-end) sequence.
If requested, the chimeric detection will be performed on the merged-mate sequence, thus allowing chimeric detection in the overlap region.
If the score of this alignment higher than the original one, or if a chimeric alignment is found, STAR will report the merged-mate aligment instead of the original one.
In the output, the merged-mate aligment will be converted back to paired-end format.
The developmment of this algorithm was supported by Illumina, Inc.
Many thanks to June Snedecor, Xiao Chen, and Felix Schlesinger for their extensive help in developing this feature.
\section{Detection of personal variants overlapping alignments.}
Option \opt{varVCFfile}\optvr{/path/to/vcf/file} is used to input VCF file with personal variants. Only single nucleotide variants (SNVs) are supported at the moment.
Each variant is expected to have a genotype with two alleles.
To output variants that overlap alignments, vG and vA have to be added to \opt{outSAMattributes} list.
SAM attribute vG outputs the genomic coordinate of the variant, allowing for identification of the variant.
SAM attribute vA outputs which allele is detected in the read: $1$ or $2$ match one of the genotype alleles, $3$ - no match to genotype.
\section{WASP filtering of allele specific alignments.}
This is re-implementation of the original WASP algorithm by Bryce van de Geijn, Graham McVicker, Yoav Gilad and Jonathan K Pritchard. Please cite the original WASP paper: Nature Methods 12, 1061–1063 (2015) \url{https://www.nature.com/articles/nmeth.3582}.
WASP filtering is activated with \opt{waspOutputMode}\optv{SAMtag}, which will add \optv{vW} tag to the SAM output:
\optv{vW:i:1} means alignment passed WASP filtering, and all other values mean it did not pass:
\optv{vW:i:2} - multi-mapping read
\optv{vW:i:3} - variant base in the read is N (non-ACGT)
\optv{vW:i:4} - remapped read did not map
\optv{vW:i:5} - remapped read multi-maps
\optv{vW:i:6} - remapped read maps to a different locus
\optv{vW:i:7} - read overlaps too many variants
\section{Detection of multimapping chimeras.}
Previous STAR chimeric detection algorithm only detected uniquely mapping chimeras, which reduced its sensitivity in some cases.
The new algorithm can detect and output multimapping chimeras. Presently, the only output into Chimeric.out.junction is supported.
This algorithm is activated with $>0$ value in \optv{chimMultimapNmax}, which defines the maximum number of chimeric multi-alignments.
The \optv{chimMultimapScoreRange} ($=1$ by default) parameter defines the score range for multi-mapping chimeras below the best chimeric score, similar to the \optv{outFilterMultimapScoreRange} parameter for normal alignments.
The \optv{chimNonchimScoreDropMin} ($=20$ by default) defines the threshold triggering chimeric detection: the drop in the best non-chimeric alignment score with respect to the read length has to be smaller than this value.
\section{Description of all options.}\label{Description_of_all_options}
For each STAR version, the most up-to-date information about all STAR parameters can be found in the \code{parametersDefault} file in the STAR source directory. The parameters in the \code{parametersDefault}, as well as in the descriptions below, are grouped by function:
