backup
This commit is contained in:
Heng Li
parent
ce859dbe1c
commit
19e4b2aab0
|
|
@ -195,3 +195,35 @@
|
|||
Title = {Optimal sequence alignment using affine gap costs},
|
||||
Volume = {48},
|
||||
Year = {1986}}
|
||||
|
||||
@article{Wu:2005vn,
|
||||
Author = {Wu, Thomas D and Watanabe, Colin K},
|
||||
Journal = {Bioinformatics},
|
||||
Pages = {1859-75},
|
||||
Title = {{GMAP}: a genomic mapping and alignment program for {mRNA} and {EST} sequences},
|
||||
Volume = {21},
|
||||
Year = {2005}}
|
||||
|
||||
@article{Iwata:2012aa,
|
||||
Author = {Iwata, Hiroaki and Gotoh, Osamu},
|
||||
Journal = {Nucleic Acids Res},
|
||||
Pages = {e161},
|
||||
Title = {Benchmarking spliced alignment programs including {Spaln2}, an extended version of {Spaln} that incorporates additional species-specific features},
|
||||
Volume = {40},
|
||||
Year = {2012}}
|
||||
|
||||
@article{Dobin:2013kx,
|
||||
Author = {Dobin, Alexander and others},
|
||||
Journal = {Bioinformatics},
|
||||
Pages = {15-21},
|
||||
Title = {{STAR}: ultrafast universal {RNA-seq} aligner},
|
||||
Volume = {29},
|
||||
Year = {2013}}
|
||||
|
||||
@article{Byrne:2017aa,
|
||||
Author = {Byrne, Ashley and others},
|
||||
Journal = {Nat Commun},
|
||||
Pages = {16027},
|
||||
Title = {Nanopore long-read {RNAseq} reveals widespread transcriptional variation among the surface receptors of individual {B} cells},
|
||||
Volume = {8},
|
||||
Year = {2017}}
|
||||
|
|
|
|||
|
|
@ -316,6 +316,9 @@ alignment.
|
|||
\end{methods}
|
||||
|
||||
\section{Results}
|
||||
|
||||
\subsection{Aligning genomic reads}
|
||||
|
||||
\begin{figure}[!tb]
|
||||
\centering
|
||||
\includegraphics[width=.5\textwidth]{roc-color.pdf}
|
||||
|
|
@ -358,6 +361,56 @@ $\ge$100bp INDELs in IGV~\citep{Robinson:2011aa} and can confirm the
|
|||
observation by~\citet{Sedlazeck169557} that BWA-MEM often breaks them into
|
||||
shorter gaps. Minimap2 does not have this issue.
|
||||
|
||||
\subsection{Aligning spliced reads}
|
||||
|
||||
\begin{table}[!tb]
|
||||
\processtable{Exon-level evaluation of 2D ONT reads from mouse}
|
||||
{\footnotesize\label{tab:exon}
|
||||
\begin{tabular}{p{3.1cm}rrrr}
|
||||
\toprule
|
||||
& GMAP & minimap2 & SpAln & STAR\\
|
||||
\midrule
|
||||
Run time (CPU min) & 631 & 15.5 & 2\,076 & 33.9 \\
|
||||
Peak RAM (GByte) & 8.9 & 14.5 & 3.2 & 29.2\vspace{1em}\\
|
||||
\# aligned reads & 103\,669 & 103\,917 & 103\,711 & 26\,479\\
|
||||
\# chimeric alignments & 1\,904 & 1\,671 & 0 & 0\\
|
||||
\# non-spliced alignments & 15\,854 & 14\,483 & 17\,033 & 10\,545\vspace{1em}\\
|
||||
\# aligned introns & 692\,275 & 694\,237 & 692\,945 & 78\,603 \\
|
||||
\# novel introns & 11\,239 & 3\,217 & 8\,550 & 1\,214 \\
|
||||
\% exact introns & 83.8\% & 91.8\% & 87.9\% & 55.2\% \\
|
||||
\% approx. introns & 91.8\% & 96.5\% & 92.5\% & 82.4\% \\
|
||||
\botrule
|
||||
\end{tabular}
|
||||
}{Reads (AC:SRR5286960) were mapped to the primary assembly of mouse genome
|
||||
GRCm38 with the following tools and command options: minimap2 (`-ax splice');
|
||||
GMAP (`-n 0 --min-intronlength 30 --cross-species'); SpAln (`-Q7 -LS -S3');
|
||||
STARlong (according to
|
||||
\href{http://bit.ly/star-pb}{http://bit.ly/star-pb}). The alignments were
|
||||
compared to the EnsEMBL gene annotation, release 89. A predicted intron
|
||||
is \emph{novel} if it has no overlaps with any annotated introns. An intron
|
||||
is \emph{exact} if it is identical to an annotated intron. An intron is
|
||||
\emph{approximate} if both of its 5'- and 3'-end are within 10bp around an
|
||||
annotated intron.}
|
||||
\end{table}
|
||||
|
||||
We evaluated minimap2 along with GMAP~(v2017-06-20; \citealp{Wu:2005vn}),
|
||||
SpAln~(v2.3.1; \citealp{Iwata:2012aa}) and STAR~(v2.5.3a;
|
||||
\citealp{Dobin:2013kx}) on real RNA-seq reads~\citep{Byrne:2017aa}.
|
||||
In general, minimap2 is more consistent with existing annotations
|
||||
(Table~\ref{tab:exon}). It finds more annotated spliced exons and predicts
|
||||
fewer novel exons. Most novel exons identified by GMAP and SpAln are
|
||||
very short, partly because the two aligners implement special routines to
|
||||
identify micro-exons. It should be possible to optimize GMAP and SpAln on this
|
||||
data set to reduce such errors. On run time, minimap2 is over 40 times faster
|
||||
than GMAP and SpAln. While STAR is close to minimap2 in speed, it does not work
|
||||
well with noisy reads.
|
||||
|
||||
We have also run aligners on the SIRV spkie-in control data (AC:SRR5286959;
|
||||
\citealp{Byrne:2017aa}) where the truth is know. Minimap2 is still the most
|
||||
accurate. 91.9\% of internal exons in the minimap2 alignment are exact.
|
||||
The percentage increases to 97.4\% if we allow up to 10bp around the splicing
|
||||
boundaries. The difference between the two percentage is mostly caused by
|
||||
|
||||
\section{Discussions}
|
||||
|
||||
Minialign and minimap2 are fast because a) with chaining, they can quickly
|
||||
|
|
|
|||
Loading…
Reference in New Issue