Test-drive the GATK tools and Best Practices pipelines on Terra


Check out this blog post to learn how you can get started with GATK and try out the pipelines in preconfigured workspaces (with a user-friendly interface!) without having to install anything.

RealignerTargetCreator and IndelRealigner

Hi,

The tools RealignerTargetCreator and IndelRealigner are obsolete in GATK4. Are there any replacements for these tools in GATK4?

Also, there are two option -BQSR (used with PrintReads) and -stand_emit_conf (used with HaplotypeCaller) that are not available with GATK4. What should be done in these cases?

Answers

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    The indel realignment tools have been retired from the best practices; they are unnecessary if you are using an assembly based caller like Mutect2 or HaplotypeCaller. Someone has proposed a port of those tools but we have not yet fully evaluated it.

    PrintReads -BQSR has been replaced by another tool called ApplyBQSR.

    The “emit” confidence threshold has been removed because it was not a principled way to annotate variant quality.

  • So the realignment steps with RealignerTargetCreator and IndelRealigner can just be deleted if using HaplotypeCaller?

    The -BQSR option was used with various tools such as PrintReads, BaseRecalibrator and AnalyzeCovariates. Does that mean that all these tools can be replaced with ApplyBQSR?

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Yes the indel realignment step can simply be skipped.

    For BQSR, no, the new tool just replaces PrintReads for that particular use case. The other two tools still exist; BaseRecalibrator is no longer able to do on the fly recalibration, so you have to run it on the recalibrated bam to do the QC cycle, and AnalyzeCovariates is used the same way as before except the argument is now lowercased (-bqsr).

  • Just to confirm, by "just replaces PrintReads for that particular use case", you mean that instead of using '-bqsr' with PrintReads, one can just use ApplyBQSR. For other PrintReads functions, we can still continue using PrintReads?

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    That's right -- for other classic PrintReads usages (eg filtering reads out of a bam, making a snippet, etc) you can still use PrintReads.

  • Are these options obsolete in VariantsToTable tool?
    "--allowMissingData", "--unsafe", "LENIENT_VCF_PROCESSING"

  • alexyfyfalexyfyf Member
    edited February 2018

    Follow the question. Now for GATK4, the best practice excludes the Realignment step.
    So are you still recommending doing BQSR since it is a downstream step of realignment in the past workflow? Also according to the best practice, the effect is marginal.
    And if you still suggest doing BQSR, do we directly feed the -BaseRecalibrator with SplitNTrim bam files (without realignment)? Am I correct?
    And is it also applicable to GATK3, skipping realignment and doing BQSR directly after SplitNTrim bam files?
    Thank you very much.

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    @Vrushali --allowMissingData is no longer necessary because the behavior wrt missing data has been flipped; prior to v. 37 the tool errored on missing data; now it emits NAs but can be told to error on missing data. The --unsafe mode switch has been removed; LENIENT_VCF_PROCESSING was replaced by --lenient.

    @alexyfyf BQSR and indel realignment have always been independent of each other; removing one does not affect the other. It also does not affect SplitNTrim. This applies equally to GATK3 and GATK4.

  • @Geraldine_VdAuwera said:
    @Vrushali --allowMissingData is no longer necessary because the behavior wrt missing data has been flipped; prior to v. 37 the tool errored on missing data; now it emits NAs but can be told to error on missing data. The --unsafe mode switch has been removed; LENIENT_VCF_PROCESSING was replaced by --lenient.

    @alexyfyf BQSR and indel realignment have always been independent of each other; removing one does not affect the other. It also does not affect SplitNTrim. This applies equally to GATK3 and GATK4.

    Thank you for your reply. To make it clear, so the "new" practice for RNAseq should be SplitNTrim > BQSR > HaplotypeCaller?

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Yes that's right.

  • AzizAziz USAMember
    Hi All,
    I was using GATK for SNPcalling for long time. It was working very well without error. Recently, I found out my command line is not working any longer. I ran the old version 3.0 and also new version 4.0 but I got the same error: line 24: GenomeAnalysisTK: command not found

    My pipeline is:
    module load bioinfo
    module load GATK/4.0.4.0
    GenomeAnalysisTK -nt 23 -T RealignerTargetCreator -R branch_gene.fasta -I JHIN01_WG_aligned_dedup.bam -o forIndelRealigner_JHIN01.intervals
    GenomeAnalysisTK -T IndelRealigner -R branch_gene.fasta -I JHIN01_WG_aligned_dedup.bam -targetIntervals forIndelRealigner_JHIN01.intervals -o sorted_JHIN01_WG_aligned.bam
    GenomeAnalysisTK -T HaplotypeCaller -R branch_gene.fasta -I sorted_JHIN01_WG_aligned.bam --emitRefConfidence GVCF -variant_index_type LINEAR -variant_index_parameter 128000 -nct 23 -o variants_JHIN01_WG.g.vcf

    I highly appreciate your help.

    Aziz
  • bhanuGandhambhanuGandham Cambridge MAMember, Administrator, Broadie, Moderator admin

    HI @Aziz

    You are not using the command line correctly for GATK4. Take a look at this doc: https://software.broadinstitute.org/gatk/documentation/quickstart

Sign In or Register to comment.