[coryfunk]

I think I figured it out.

The indices were created with an older version of STAR, such that when I tried running the newest version of STAR on the old indices, nothing worked.

[coryfunk]

Alex,

I'ved used older versions of STAR before and I'm now trying to use the latest 2.4.2a. I'd like to try the --quantMode GeneCounts option and I'm having trouble. From the documentation, as long as I included the -sjdbGTFfile option in building my indices, I should be able to use the --quantMode option. Here's how I built my indices:

STAR --runMode genomeGenerate --genomeDir /mnt/indices --genomeFastaFiles /mnt/genome/Homo_sapiens.GRCh38.dna.SORTED.fa --sjdbGTFfile /mnt/transcriptome/Homo_sapiens.GRCh38.77.gtf --runThreadN 32

Here's how I'm trying to run STAR:

STAR --genomeDir /mnt/indices --readFilesIn /mnt/data/NA10-080.FCC6178ACXX_L3_R1_ITAGCTT.fastq.gz /mnt/data/NA10-080.FCC6178ACXX_L3_R2_ITAGCTT.fastq.gz --readFilesCommand zcat -sjdbGTFfile /mnt/transcriptome/Homo_sapiens.GRCh38.77.gtf --quantMode TranscriptomeSAM GeneCounts --runThreadN 32 --outFilterMismatchNmax 8 --outFileNamePrefix NA10-080.star_new --outFilterMultimapNmax 10

I've tried changing the order of some things, but I keep getting the following error:

gzip: invalid option -- 's'
Try `gzip --help' for more information.
Jul 07 19:17:22 ..... Started STAR run
Jul 07 19:17:22 ..... Loading genome
gzip: invalid option -- 's'
Try `gzip --help' for more information.

ERROR_011001: exiting because of *INPUT FILE* error: could not open input file /mnt/indices/transcriptInfo.tab
Solution: check that the file exists and you have read permission for this file
SOLUTION: utilize --sjdbGTFfile /path/to/annotantions.gtf option at the genome generation step or mapping step


At one point it looked like it wasn't correctly pointing to the transcriptInfo.tab file because of how I specified this: --genomeDir /mnt/indices/ giving me the following error:

ERROR_011001: exiting because of *INPUT FILE* error: could not open input file /mnt/indices//transcriptInfo.tab
Solution: check that the file exists and you have read permission for this file
SOLUTION: utilize --sjdbGTFfile /path/to/annotantions.gtf option at the genome generation step or mapping step

I took off the last "/" and it appeared to alter how it assembled the path (see first error message above) but still doesn't seem to find the file.

Thoughts?

[alexdobin]

Dear Alex,

i started to use STAR and am really happy with it.

There are just a few questions:

1. We have different settings. In one we have 150 bases single end reads
in the other 75 bases paired end reads. I am using a gtf file with annotated splice junctions. What would be the ideal value for --sjdbOverhang in the single end case be (149 ?) or would the general value of 100 should not be fine and how does read trimming influence this choice ? As then the read lengths would vary.
Why does it even make a difference if i have a read length of 150 and an overhang of 150 there would be no splicing anyway. So the values for --sjdbOverhang ould just have to be higher than the maximum read length ?

2. When using a gtf annotation and i use the filter BySJout option
does it remove all reads that that contain non annotated splices. Because in the log file it says so.

3. Is there a best practice way to filter unreliable novel junctions before the two step procedure.
And does the two step procedure hav an influence on differential expression for well annotated genomes
such as arabidopsis.

4. We plan to have many measurements over a long time, what would be the
ideal setting of STAR to make sure that the values are comparable for differential expression.

5. What happens when only one read of a read was mapped are both discarded ?
Do both reads always have the same mapping quality ?

May thanks for your help.

Hi @IonTom,

good questions, here are my answers:

1. The --sjdbOvehang=100 (which is set by default in the latest versions of STAR) is fine in most of the applications. Users can still fine tune it, with impact being higher for short reads (<50b). These thread , thread discuss the workings of --sjdbOverhang in more detail.

2. --outFilterType BySJout option will keep reads splicing across the annotated junctions, as well as novel junctions that pass filtering by --outSJfilter* options (only those filtered junctions that are output in SJ.out.tab). In other words, it synchronizes the alignments (SAM or BAM) with the SJ.out.tab file. --outSJfilter* filters include filtering by the minimum splice ovehang, number of reads per junction, intron length, and closeness to other junctions.

3. Since the junctions for the 2-pass mapping are taken from the SJ.out.tab file, they are already filtered with some default values of the --outSJfilter* filters. You can review this filtering to see if you want to make it more or less stringent. It is advisable to remove junctions from those references that are not supposed to splice, such as chrM or spike-ins, these are false-positives and may slow the 2nd pass mapping.
Also, if you get a large number of non-canonical junctions, it may be helpful to filter them more stringently.

4. To keep things comparable over long time on the pipeline side, you need to keep the same run parameters, genome, and annotations. STAR versions/patches typically produce exactly the same results, unless specific changes in the mapping algorithm are announced.

5. By default, only correctly paired alignments are output. It is possible (but not recommended) to output single-end or non-concordant pairs.
The mapping quality of two reads in a pair is not required to be the same, on the contrary, the mapping matrics of the whole pair (alignment score, length, number of mismatches, etc) are controlled.

Cheers
Alex

[IonTom]

Dear Alex,

i started to use STAR and am really happy with it.

There are just a few questions:

1. We have different settings. In one we have 150 bases single end reads
in the other 75 bases paired end reads. I am using a gtf file with annotated splice junctions. What would be the ideal value for --sjdbOverhang in the single end case be (149 ?) or would the general value of 100 should not be fine and how does read trimming influence this choice ? As then the read lengths would vary.
Why does it even make a difference if i have a read length of 150 and an overhang of 150 there would be no splicing anyway. So the values for --sjdbOverhang ould just have to be higher than the maximum read length ?

2. When using a gtf annotation and i use the filter BySJout option
does it remove all reads that that contain non annotated splices. Because in the log file it says so.

3. Is there a best practice way to filter unreliable novel junctions before the two step procedure.
And does the two step procedure hav an influence on differential expression for well annotated genomes
such as arabidopsis.

4. We plan to have many measurements over a long time, what would be the
ideal setting of STAR to make sure that the values are comparable for differential expression.

5. What happens when only one read of a read was mapped are both discarded ?
Do both reads always have the same mapping quality ?



May thanks for your help.

[neokao]

I did : /usr/local/Cellar/gcc49/bin/g++-4.9 -v and got this:
-bash: /usr/local/Cellar/gcc49/bin/g++-4.9: No such file or directory.

Anyway, I downloaded your 2.4.0k Mac executable and it worked fine with that particular .fq file now. Thanks a lot for all the patience helping me.

[alexdobin]

i switched to source directory.
grep STARforMacStatic Makefile
STARforMacStatic : CCFLAGS=-D'COMPILE_FOR_MAC' -I ./Mac_Include/ $(CCFLAGS_main)
STARforMacStatic : parametersDefault.xxd $(OBJECTS)

Then I got this error:

make STARforMacStatic CXX=/usr/local/Cellar/gcc49/bin/g++-4.9
/usr/local/Cellar/gcc49/bin/g++-4.9 -o STAR -D'COMPILE_FOR_MAC' -I ./Mac_Include/ -O3 -pipe -std=c++0x -Wall -Wextra -fopenmp -D'COMPILATION_TIME_PLACE="Wed Apr 8 12:34:50 CST 2015 :/Users/Matrix_Macmini/NGS/STAR-STAR_2.4.0k/source"' -pthread -lz -static-libgcc htslib/libhts.a PackedArray.o SuffixArrayFuns.o STAR.o Parameters.o InOutStreams.o SequenceFuns.o Genome.o ...
make: /usr/local/Cellar/gcc49/bin/g++-4.9: No such file or directory
make: *** [STARforMacStatic] Error 1

It looks like you path to /usr/local/Cellar/gcc49/bin/g++-4.9 is broken. Can you run /usr/local/Cellar/gcc49/bin/g++-4.9 -v ?

I have posted the 2.4.0k Mac executable here:

http://labshare.cshl.edu/shares/gingeraslab/www-data/dobin/STAR/STARreleases/2.4.0k/bin/MacOSX_x86_64/STAR

Hopefully, it will work on your system and save you the compiling troubles.

Cheers
Alex

[neokao]

i switched to source directory.
grep STARforMacStatic Makefile
STARforMacStatic : CCFLAGS=-D'COMPILE_FOR_MAC' -I ./Mac_Include/ $(CCFLAGS_main)
STARforMacStatic : parametersDefault.xxd $(OBJECTS)

Then I got this error:

make STARforMacStatic CXX=/usr/local/Cellar/gcc49/bin/g++-4.9
/usr/local/Cellar/gcc49/bin/g++-4.9 -o STAR -D'COMPILE_FOR_MAC' -I ./Mac_Include/ -O3 -pipe -std=c++0x -Wall -Wextra -fopenmp -D'COMPILATION_TIME_PLACE="Wed Apr 8 12:34:50 CST 2015 :/Users/Matrix_Macmini/NGS/STAR-STAR_2.4.0k/source"' -pthread -lz -static-libgcc htslib/libhts.a PackedArray.o SuffixArrayFuns.o STAR.o Parameters.o InOutStreams.o SequenceFuns.o Genome.o Stats.o Transcript.o Transcript_alignScore.o ReadAlign.o ReadAlign_storeAligns.o ReadAlign_stitchPieces.o ReadAlign_multMapSelect.o ReadAlign_mapOneRead.o readLoad.o ReadAlignChunk.o ReadAlignChunk_processChunks.o ReadAlignChunk_mapChunk.o OutSJ.o outputSJ.o blocksOverlap.o ThreadControl.o sysRemoveDir.o ReadAlign_maxMappableLength2strands.o binarySearch2.o ReadAlign_outputAlignments.o ReadAlign_outputTranscriptSAM.o ReadAlign_outputTranscriptSJ.o ReadAlign_outputTranscriptCIGARp.o ReadAlign_createExtendWindowsWithAlign.o ReadAlign_assignAlignToWindow.o ReadAlign_oneRead.o ReadAlign_stitchWindowSeeds.o ReadAlign_chimericDetection.o stitchWindowAligns.o extendAlign.o stitchAlignToTranscript.o alignSmithWaterman.o genomeGenerate.o TimeFunctions.o ErrorWarning.o loadGTF.o streamFuns.o stringSubstituteAll.o Transcriptome.o Transcriptome_quantAlign.o ReadAlign_quantTranscriptome.o Quantifications.o Transcriptome_geneCountsAddAlign.o sjdbLoadFromFiles.o sjdbLoadFromStream.o sjdbPrepare.o sjdbBuildIndex.o mapThreadsSpawn.o Parameters_openReadsFiles.cpp Parameters_closeReadsFiles.cpp BAMoutput.o BAMfunctions.o ReadAlign_alignBAM.o BAMbinSortByCoordinate.o signalFromBAM.o bamRemoveDuplicates.o BAMbinSortUnmapped.o bam_cat.o
make: /usr/local/Cellar/gcc49/bin/g++-4.9: No such file or directory
make: *** [STARforMacStatic] Error 1

[alexdobin]

I used homebrew to install gcc again. The installed path is /usr/local/Cellar/gcc49.

Then I did make STARforMacStatic CXX=/usr/local/Cellar/gcc49/bin/g++-4.9 and I got another error:
make: *** No rule to make target `STARforMacStatic'. Stop.

Really frustrating now.

Are you running make from the STAR source directory? This error often occurs if make is run from a wrong directory.
If you do:
grep STARforMacStatic Makefile
it will show you the STARforMacStatic target in the Makefile:
STARforMacStatic : CCFLAGS=-D'COMPILE_FOR_MAC' -I ./Mac_Include/ $(CCFLAGS_main)
STARforMacStatic : parametersDefault.xxd $(OBJECTS)


Cheers
Alex

[neokao]

I used homebrew to install gcc again. The installed path is /usr/local/Cellar/gcc49.

Then I did make STARforMacStatic CXX=/usr/local/Cellar/gcc49/bin/g++-4.9 and I got another error:
make: *** No rule to make target `STARforMacStatic'. Stop.

Really frustrating now.

[alexdobin]

The Mac's /usr/bin/gcc (which is on the PATH so you can invoke it with simply gcc) symlinks to clang, so you are still trying to compile with clang.

When you installed (configured) true gcc, did you use --prefix option? You need to find installation path for the true gcc.
You need to be able to check the version:
/path/to/gcc/g++ -v

and it should say something like (not Apple LLVM clang etc):

Using built-in specs.
Target: x86_64-redhat-linux
Configured with: ../configure --prefix=/opt/hpc
Thread model: posix
gcc version 4.4.6 20110731 (Red Hat 4.4.6-3) (GCC)

If it works, compile STAR with (it has to be g++, not gcc - I made a mistake in the previous post):
make STARforMacStatic CXX=/path/to/g++

Cheers
Alex

[GenoMax]

See if the second answer in this thread helps:http://stackoverflow.com/questions/1...-osx-mavericks

Otherwise you will have to wait for Alex to respond.

[neokao]

Yes. My gcc is there.
So I did make STARforMacStatic CXX=/usr/bin/gcc but still got error:

clang: error: unsupported option '-static-libgcc'
make: *** [STARforMacStatic] Error 1

Thanks folks.

Your gcc should be in /usr/bin. Confirm that by

$ which gcc

When you compile STAR do

$ make STARforMacStatic CXX=/usr/bin/gcc

[GenoMax]

Your gcc should be in /usr/bin. Confirm that by

$ which gcc

When you compile STAR do

$ make STARforMacStatic CXX=/usr/bin/gcc

[neokao]

Compilation on Mac is tricky because the default compiler - clang - does not support OMP used by STAR. Please try to compile with
make STARforMacStatic CXX=/path/to/gcc

Following your suggestion, I got a different error:
/bin/sh: /path/to/gcc: No such file or directory
make: *** [Depend.list] Error 127

I did install gcc.
gcc --version
Configured with: --prefix=/Applications/Xcode.app/Contents/Developer/usr --with-gxx-include-dir=/usr/include/c++/4.2.1
Apple LLVM version 6.0 (clang-600.0.57) (based on LLVM 3.5svn)
Target: x86_64-apple-darwin14.1.0
Thread model: posix

Thanks.

[alexdobin]

Compilation on Mac is tricky because the default compiler - clang - does not support OMP used by STAR. Please try to compile with
make STARforMacStatic CXX=/path/to/gcc