Ready for 2.7.9a

alexdobin · May 4, 2021 · 1ebfe0b · 1ebfe0b
1 parent 4b66c5e
commit 1ebfe0b
Show file tree

Hide file tree

Showing 9 changed files with 60 additions and 9 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -1,8 +1,8 @@
-STAR 2.7.9a --- 2021/02/20 ::: STARsolo updates
+STAR 2.7.9a --- 2021/05/05 ::: STARsolo updates
 =====================================================
 **Major updates:**
 * STARsolo can perform counting of multi-gene (multi-mapping) reads with --soloMultiMappers EM [Uniform Rescue PropUnqiue] options.
-* PR #1163: SIMDe takes care of correct SIMD extensions based on -m g++ flag: compilation option CXXFLAGS_SIMD is preset to -mavx2, but can be to the desired target architecture. Many thanks to Michael R. Crusoe @mr-c, Evan Nemerson @nemequ and Steffen Möller @smoe!
+* PR #1163: [SIMDe](https://github.com/simd-everywhere/simde) takes care of correct SIMD extensions based on -m g++ flag: compilation option CXXFLAGS_SIMD is preset to -mavx2, but can be to the desired target architecture. Many thanks to Michael R. Crusoe @mr-c, Evan Nemerson @nemequ and Steffen Möller @smoe!
 
 **New options and features:**
 * New option: --soloUMIfiltering MultiGeneUMI_All to filter out all UMIs mapping to multiple genes (for uniquely mapping reads)

diff --git a/README.md b/README.md
@@ -1,5 +1,5 @@
-STAR 2.7
-========
+STAR 2.7.9a
+==========
 Spliced Transcripts Alignment to a Reference
 © Alexander Dobin, 2009-2021
 https://www.ncbi.nlm.nih.gov/pubmed/23104886
@@ -52,6 +52,11 @@ Compile under Linux
 cd STAR/source
 make STAR
 ```
+For processors that do not support AVX extensions, specify the target SIMD architecture, e.g.
+```
+make STAR CXXFLAGS_SIMD=sse
+```
+
 
 Compile under Mac OS X
 ----------------------

diff --git a/RELEASEnotes.md b/RELEASEnotes.md
@@ -1,3 +1,13 @@
+STAR 2.7.9a --- 2021/05/05 ::: STARsolo updates
+=====================================================
+
+* [**Counting *multi-gene* (multimapping) reads**](#multi-gene-reads)
+* STARsolo uses [SIMDe](https://github.com/simd-everywhere/simde) package which support different types of SIMD extensions. For processors that do not support AVX extensions, specify the target SIMD architecture, e.g.
+```
+make STAR CXXFLAGS_SIMD=sse
+```
+
+
 STAR 2.7.8a --- 2021/02/20
 ===========================
 **Major STARsolo updates and many bug fixes**

diff --git a/bin/Linux_x86_64/STAR b/bin/Linux_x86_64/STAR
diff --git a/bin/Linux_x86_64/STARlong b/bin/Linux_x86_64/STARlong
diff --git a/bin/Linux_x86_64_static/STAR b/bin/Linux_x86_64_static/STAR
diff --git a/bin/Linux_x86_64_static/STARlong b/bin/Linux_x86_64_static/STARlong
diff --git a/docs/STARsolo.md b/docs/STARsolo.md
@@ -1,6 +1,10 @@
 **STARsolo**: mapping, demultiplexing and quantification for single cell RNA-seq
 =================================================================================
 
+Major update in STAR 2.7.9a (2021/05/05)
+-----------------------------------------
+* [**Counting *multi-gene* (multimapping) reads**](#multi-gene-reads)
+
 Major updates in STAR 2.7.8a (2021/02/20)
 -----------------------------------------
 * [**Cell calling (filtering) similar to CellRanger:**](#emptydrop-like-filtering)
@@ -181,7 +185,7 @@ int>0:                  maximum number of mismatches allowed in adapter sequence
 
 --------------------------------------
 Cell filtering (calling)
---------------------
+--------------------------------------
 In addition to raw, unfiltered output of gene/cell counts, STARsolo performs cell filtering (a.k.a. cell calling), which aims to select a subset of cells that are likely to be "real" cells as opposed to empty droplets (containing ambient RNA).
 Two types of filtering are presently implemented: simple (knee-like) and advanced EmptyDrop-like. The selected filtering is also used to produce summary statistics for filtered cells in the Summary.csv file, which is similar to CellRanger's summary and is useful for Quality Control.
 
@@ -230,9 +234,40 @@ Quantification of different transcriptomic features
         --soloFeatures Gene GeneFull SJ Velocyto
         ```
 
+------------------------------------------------------
+Multi-gene reads
+------------------------------------------------------        
+Multi-gene reads are concordant with (i.e. align equally well to) transcripts of two or more genes. One class of multi-gene read are those that map uniquely to a genomic region where two or more genes overlap. Another class are those reads that map to multiple loci in the genome, with each locus annotated to a different gene.
+
+Including multi-gene reads allows for more accurate gene quantification and, more importantly, enables detection of gene expression from certain classes of genes that are supported only by multi-gene reads, such as overlapping genes and highly similar paralog families.
+
+The multi-gene read recovery options are specified with ```--soloMultiMappers```. Several algorithms are implemented:
+
+```
+--soloMultiMappers Uniform
+```
+uniformly distributes the multi-gene UMIs to all genes in its gene set. Each gene gets a fractional count of 1/N_genes, where N_genes is the number of genes in the set. This is the simplest possible option, and it offers higher sensitivity for gene detection at the expense of lower precision.
+
+```
+--soloMultiMappers Uniform
+```
+ distributes the multi-gene UMIs proportionally to the number of unqiue UMIs per gene. UMIs that map to genes that are not supported by unique UMIs are distributed uniformly.
+
+```
+--soloMultiMappers EM
+```
+uses Maximum Likelihood Estimation (MLE) to distribute multi-gene UMIs among their genes, taking into account other UMIs (both unique- and multi-gene) from the same cell (i.e. with the same CB). Expectation-Maximization (EM) algorithm is used to find the gene expression values that maximize the likelihood function. Recovering multi-gene reads via MLE-EM model was previously used to quantify transposable elements in bulk RNA-seq {[TEtranscripts](https://doi.org/10.1093/bioinformatics/btv422)} and in scRNA-seq {[Alevin](https://doi.org/10.1186/s13059-019-1670-y); [Kallisto-bustools](http://www.nature.com/articles/s41587-021-00870-2)}.
+
+```
+--soloMultiMappers Rescue
+```
+distributes multi-gene UMIs to their gene set proportionally to the sum of the number of unique-gene UMIs and uniformly distributed multi-gene UMIs in each gene [Mortazavi et al](https://www.nature.com/articles/nmeth.1226). It can be thought of as the first step of the EM algorithm.
+
+Any combination of these options can be specified and different multi-gene falvors will be output into different files. The unique-gene UMI counts are output into the *matrix.mtx* file in the *raw/Gene* directory, while the sum of unique+multi-gene UMI counts will be output into *UniqueAndMult-EM.mtx,  UniqueAndMult-PropUnique.mtx,  UniqueAndMult-Rescue.mtx,  UniqueAndMult-Uniform.mtx* files.
+
 --------------------------------------
 BAM tags
------------------
+--------------------------------------
 * To output BAM tags into SAM/BAM file, add them to the list of standard tags in
     ```
     --outSAMattributes NH HI nM AS CR UR CB UB GX GN sS sQ sM

diff --git a/source/parametersDefault.xxd b/source/parametersDefault.xxd
@@ -1246,8 +1246,9 @@ unsigned char parametersDefault[] = {
   0x43, 0x6f, 0x6f, 0x72, 0x64, 0x69, 0x6e, 0x61, 0x74, 0x65, 0x20, 0x2e,
   0x2e, 0x2e, 0x20, 0x61, 0x6c, 0x69, 0x67, 0x6e, 0x6d, 0x65, 0x6e, 0x74,
   0x73, 0x20, 0x69, 0x6e, 0x20, 0x42, 0x41, 0x4d, 0x20, 0x66, 0x6f, 0x72,
-  0x6d, 0x61, 0x74, 0x2c, 0x20, 0x75, 0x6e, 0x73, 0x6f, 0x72, 0x74, 0x65,
-  0x64, 0x2e, 0x20, 0x52, 0x65, 0x71, 0x75, 0x69, 0x72, 0x65, 0x73, 0x20,
+  0x6d, 0x61, 0x74, 0x2c, 0x20, 0x73, 0x6f, 0x72, 0x74, 0x65, 0x64, 0x20,
+  0x62, 0x79, 0x20, 0x63, 0x6f, 0x6f, 0x72, 0x64, 0x69, 0x6e, 0x61, 0x74,
+  0x65, 0x2e, 0x20, 0x52, 0x65, 0x71, 0x75, 0x69, 0x72, 0x65, 0x73, 0x20,
   0x2d, 0x2d, 0x6f, 0x75, 0x74, 0x53, 0x41, 0x4d, 0x74, 0x79, 0x70, 0x65,
   0x20, 0x42, 0x41, 0x4d, 0x20, 0x53, 0x6f, 0x72, 0x74, 0x65, 0x64, 0x42,
   0x79, 0x43, 0x6f, 0x6f, 0x72, 0x64, 0x69, 0x6e, 0x61, 0x74, 0x65, 0x0a,
@@ -4469,4 +4470,4 @@ unsigned char parametersDefault[] = {
   0x65, 0x76, 0x65, 0x6c, 0x6f, 0x70, 0x6d, 0x65, 0x6e, 0x74, 0x5f, 0x65,
   0x6e, 0x64, 0x0a
 };
-unsigned int parametersDefault_len = 53619;
+unsigned int parametersDefault_len = 53631;