Recent changes to LoFreq-Star-FAQ

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Mon, 05 May 2014 05:04:22 -0000

--- v10
+++ v11
@@ -1,4 +1,4 @@
-# FAQ for LoFreq* 
+# FAQ for LoFreq-Star

 ##### How to Cite

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Mon, 05 May 2014 05:02:53 -0000

--- v9
+++ v10
@@ -1,4 +1,4 @@
-# UNDER CONSTRUCTION: FAQ for LoFreq* 
+# FAQ for LoFreq*

 ##### How to Cite

@@ -11,7 +11,7 @@

 ##### Do I need to filter LoFreq Predictions?

-You usually don't. Predicted variants are already filtered using default parameters (which include coverage, strand-bias, snv-quality etc). If you however, need extra conservative settings you can use `lofreq filter` to add some more filter criteria
+You usually don't. Predicted variants are already filtered using default parameters (which include coverage, strand-bias, snv-quality etc). If you however, need extra conservative settings you can use `lofreq filter` to add some more filter criteria.

 ##### What Happens if I didn't recalibrate Base-Qualities?

@@ -22,17 +22,13 @@
 If you want to run LoFreq only on certain regions use the appropriate bed-file as input with `-l regions.bed`. Not doing so will negatively affect LoFreq sensitivity and it might calls variants
 outside of the desired regions. 

-
 ##### PCR Amplified Data

-If your data was heavily PCR-amplified LoFreq will likely call SNVs in the primer regions as well, due
+If your data was heavily PCR-amplified LoFreq will likely call SNVs in primer regions, due
 to ambiguous primer positions, primer impurities etc. You should ignore primer positions after SNV
-calling. The best way to achieve this is to create a bed-file that only lists non-primer regions and use that as input to Lofreq with `-l regions.bed`
+calling. The best way to achieve this is to create a bed-file that only lists non-primer regions and use that as input to Lofreq with `-l regions.bed`.

-Another problem with heavily PCRed input is that PCR artifacts might show up as low-frequency SNVs,
-especially if a mis-amplification happened in early cycles. Computational tools will be 
-unable to distinguish these from true low-frequency variants, since the mis-incorporated bases look 
-real to the sequencing machine.
+Another problem with heavily PCRed input is that PCR artifacts will show up as low-frequency SNVs. Their allele frequency will usually be low, unless a mis-amplification happened in early cycles. Computational tools will be unable to distinguish these from true low-frequency variants, since the mis-incorporated bases look real to the sequencing machine. To get rid of these you will either have to run your samples in duplicates (before amplification) or remove SNVs below the expected PCR error frequency.

 ##### I get fewer SNVs than Expected When Running LoFreq on a BAM file with Bowtie/BWA-SW

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Tue, 08 Apr 2014 05:08:23 -0000

--- v8
+++ v9
@@ -1,29 +1,29 @@
 # UNDER CONSTRUCTION: FAQ for LoFreq*

-#### How to Cite
+##### How to Cite

 Please cite: [Wilm et a. LoFreq: A sequence-quality aware, ultra-sensitive variant caller for uncovering cell-population heterogeneity from high-throughput sequencing datasets. _Nucleic Acids Res._ 2012; 40(22):11189-201.](http://www.ncbi.nlm.nih.gov/pubmed/23066108)

-#### Where to get Help:
+##### Where to get Help:

 - See the [mailing list](https://sourceforge.net/p/lofreq/mailman/)
 - See [the LoFreq blog](https://sourceforge.net/p/lofreq/blog/) for news and updates

-#### Do I need to filter LoFreq Predictions?
+##### Do I need to filter LoFreq Predictions?

 You usually don't. Predicted variants are already filtered using default parameters (which include coverage, strand-bias, snv-quality etc). If you however, need extra conservative settings you can use `lofreq filter` to add some more filter criteria

-#### What Happens if I didn't recalibrate Base-Qualities?
+##### What Happens if I didn't recalibrate Base-Qualities?

 We usually recommend to follow GATK's best practices on post-processing of you BAM file which include base quality recalibration. You can run LoFreq on non-recalibrated data, which might however results in a higher rate of false positive calls. We would recommend GATK base-recalibration even for non-human data or targeted sequencing. GATK usually requires the input of known variant sites (which is a circular problem) which are however not known for most organisms non-human data (use dbsnp for human data). One option is to use initial LoFreq predictions as vcf input to GATK and then run LoFreq again.

-#### Regions for Targeted Sequencing
+##### Regions for Targeted Sequencing

 If you want to run LoFreq only on certain regions use the appropriate bed-file as input with `-l regions.bed`. Not doing so will negatively affect LoFreq sensitivity and it might calls variants
 outside of the desired regions. 

-#### PCR Amplified Data
+##### PCR Amplified Data

 If your data was heavily PCR-amplified LoFreq will likely call SNVs in the primer regions as well, due
 to ambiguous primer positions, primer impurities etc. You should ignore primer positions after SNV
@@ -34,10 +34,10 @@
 unable to distinguish these from true low-frequency variants, since the mis-incorporated bases look 
 real to the sequencing machine.

-### I get fewer SNVs than Expected When Running LoFreq on a BAM file with Bowtie/BWA-SW
+##### I get fewer SNVs than Expected When Running LoFreq on a BAM file with Bowtie/BWA-SW

 BWA-SW assigns very low mapping qualities to mapped reads. Consider disabling the use of mapping quality on `lofreq call` with `-J`. The same is true for Bowtie, depending on the parameters you used. Bowtie it will sometimes produce alignments with reads that have either only mapping quality 0 (non unique) or 255 (not available). It is unclear to us why that happens, or what exactly that means (see also [this thread](http://seqanswers.com/forums/showthread.php?t=3142)). LoFreq uses mapping quality by default for the prediction of SNVs, which doesn't help in this scenario. We would suggest to switch off the use of mapping quality (see above) in such cases.

-#### `Call-Parallel` Fails with an Error Message Which is Hard to Interpret
+##### `Call-Parallel` Fails with an Error Message Which is Hard to Interpret

 This is easiest to debug by running the single-threaded version `call` first with the same parameters, check the error message and fix the command accordingly. Then execute `call-parallel` again with fixed parameters.

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Tue, 08 Apr 2014 05:07:40 -0000

--- v7
+++ v8
@@ -1,29 +1,29 @@
 # UNDER CONSTRUCTION: FAQ for LoFreq*

-#### How to cite
+#### How to Cite

 Please cite: [Wilm et a. LoFreq: A sequence-quality aware, ultra-sensitive variant caller for uncovering cell-population heterogeneity from high-throughput sequencing datasets. _Nucleic Acids Res._ 2012; 40(22):11189-201.](http://www.ncbi.nlm.nih.gov/pubmed/23066108)

-#### Where to get help:
+#### Where to get Help:

 - See the [mailing list](https://sourceforge.net/p/lofreq/mailman/)
 - See [the LoFreq blog](https://sourceforge.net/p/lofreq/blog/) for news and updates

-#### Do I need to filter LoFreq predictions
+#### Do I need to filter LoFreq Predictions?

 You usually don't. Predicted variants are already filtered using default parameters (which include coverage, strand-bias, snv-quality etc). If you however, need extra conservative settings you can use `lofreq filter` to add some more filter criteria

-#### What happens if I didn't recalibrate base-qualities
+#### What Happens if I didn't recalibrate Base-Qualities?

 We usually recommend to follow GATK's best practices on post-processing of you BAM file which include base quality recalibration. You can run LoFreq on non-recalibrated data, which might however results in a higher rate of false positive calls. We would recommend GATK base-recalibration even for non-human data or targeted sequencing. GATK usually requires the input of known variant sites (which is a circular problem) which are however not known for most organisms non-human data (use dbsnp for human data). One option is to use initial LoFreq predictions as vcf input to GATK and then run LoFreq again.

-#### Regions for targeted sequencing
+#### Regions for Targeted Sequencing

 If you want to run LoFreq only on certain regions use the appropriate bed-file as input with `-l regions.bed`. Not doing so will negatively affect LoFreq sensitivity and it might calls variants
 outside of the desired regions. 

-#### PCR amplified data
+#### PCR Amplified Data

 If your data was heavily PCR-amplified LoFreq will likely call SNVs in the primer regions as well, due
 to ambiguous primer positions, primer impurities etc. You should ignore primer positions after SNV
@@ -34,14 +34,10 @@
 unable to distinguish these from true low-frequency variants, since the mis-incorporated bases look 
 real to the sequencing machine.

-### I get strange results with Bowtie
+### I get fewer SNVs than Expected When Running LoFreq on a BAM file with Bowtie/BWA-SW

-Depending on the parameters you used for calling Bowtie it will sometimes produces alignments with reads that have either only mapping quality 0 (non unique) or MQ 255 (not available). It is unclear to us why that happens, or what exactly that means (see also [this thread](http://seqanswers.com/forums/showthread.php?t=3142)). Anyway, LoFreq will by default use mapping quality for the prediction of SNVs, which doesn't help in this scenario though. If you have to use Bowtie we suggest to switch off the use of mapping quality in `lofreq call` with `-J`.
+BWA-SW assigns very low mapping qualities to mapped reads. Consider disabling the use of mapping quality on `lofreq call` with `-J`. The same is true for Bowtie, depending on the parameters you used. Bowtie it will sometimes produce alignments with reads that have either only mapping quality 0 (non unique) or 255 (not available). It is unclear to us why that happens, or what exactly that means (see also [this thread](http://seqanswers.com/forums/showthread.php?t=3142)). LoFreq uses mapping quality by default for the prediction of SNVs, which doesn't help in this scenario. We would suggest to switch off the use of mapping quality (see above) in such cases.

-### I get strange results with BWA-SW
-
-See also 'Bowtie' above. BWA-SW assigns very low mapping qualities to mapped reads. Consider disabling the use of mapping quality on `lofreq call` with `-J`.
-
-#### `call-parallel` fails with a non-descript error message
+#### `Call-Parallel` Fails with an Error Message Which is Hard to Interpret

 This is easiest to debug by running the single-threaded version `call` first with the same parameters, check the error message and fix the command accordingly. Then execute `call-parallel` again with fixed parameters.

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Tue, 08 Apr 2014 05:03:56 -0000

--- v6
+++ v7
@@ -34,10 +34,13 @@
 unable to distinguish these from true low-frequency variants, since the mis-incorporated bases look 
 real to the sequencing machine.

-### I get strange results with Bowtie. 
+### I get strange results with Bowtie

-Depending on the parameters you used for calling Bowtie it will sometimes produces alignments with reads that have either only mapping quality 0 (non unique) or MQ 255 (not available). It is unclear to us why that happens, or what exactly that means (see also [this thread](http://seqanswers.com/forums/showthread.php?t=3142)). Anyway, LoFreq will by default use mapping quality for the prediction of SNVs, which doesn't help in this scenario though. If you have to use Bowtie we suggest to switch off the use of mapping quality in LoFreq with `-J`
+Depending on the parameters you used for calling Bowtie it will sometimes produces alignments with reads that have either only mapping quality 0 (non unique) or MQ 255 (not available). It is unclear to us why that happens, or what exactly that means (see also [this thread](http://seqanswers.com/forums/showthread.php?t=3142)). Anyway, LoFreq will by default use mapping quality for the prediction of SNVs, which doesn't help in this scenario though. If you have to use Bowtie we suggest to switch off the use of mapping quality in `lofreq call` with `-J`.

+### I get strange results with BWA-SW
+
+See also 'Bowtie' above. BWA-SW assigns very low mapping qualities to mapped reads. Consider disabling the use of mapping quality on `lofreq call` with `-J`.

 #### `call-parallel` fails with a non-descript error message

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Mon, 07 Apr 2014 05:58:24 -0000

--- v5
+++ v6
@@ -2,13 +2,16 @@

 #### How to cite

-Please cite [Wilm et a. LoFreq: A sequence-quality aware, ultra-sensitive variant caller for uncovering cell-population heterogeneity from high-throughput sequencing datasets. _Nucleic Acids Res._ 2012; 40(22):11189-201.](http://www.ncbi.nlm.nih.gov/pubmed/23066108)
+Please cite: [Wilm et a. LoFreq: A sequence-quality aware, ultra-sensitive variant caller for uncovering cell-population heterogeneity from high-throughput sequencing datasets. _Nucleic Acids Res._ 2012; 40(22):11189-201.](http://www.ncbi.nlm.nih.gov/pubmed/23066108)

 #### Where to get help:

 - See the [mailing list](https://sourceforge.net/p/lofreq/mailman/)
 - See [the LoFreq blog](https://sourceforge.net/p/lofreq/blog/) for news and updates

+#### Do I need to filter LoFreq predictions
+
+You usually don't. Predicted variants are already filtered using default parameters (which include coverage, strand-bias, snv-quality etc). If you however, need extra conservative settings you can use `lofreq filter` to add some more filter criteria

 #### What happens if I didn't recalibrate base-qualities

@@ -31,7 +34,7 @@
 unable to distinguish these from true low-frequency variants, since the mis-incorporated bases look 
 real to the sequencing machine.

-### I get strange results with Bowtie. Why?
+### I get strange results with Bowtie. 

 Depending on the parameters you used for calling Bowtie it will sometimes produces alignments with reads that have either only mapping quality 0 (non unique) or MQ 255 (not available). It is unclear to us why that happens, or what exactly that means (see also [this thread](http://seqanswers.com/forums/showthread.php?t=3142)). Anyway, LoFreq will by default use mapping quality for the prediction of SNVs, which doesn't help in this scenario though. If you have to use Bowtie we suggest to switch off the use of mapping quality in LoFreq with `-J`

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Mon, 07 Apr 2014 05:56:42 -0000

--- v4
+++ v5
@@ -1,26 +1,41 @@
 # UNDER CONSTRUCTION: FAQ for LoFreq*

-- how to cite
-- where to get help: [https://sourceforge.net/p/lofreq/mailman/](mailing list)
+#### How to cite
+
+Please cite [Wilm et a. LoFreq: A sequence-quality aware, ultra-sensitive variant caller for uncovering cell-population heterogeneity from high-throughput sequencing datasets. _Nucleic Acids Res._ 2012; 40(22):11189-201.](http://www.ncbi.nlm.nih.gov/pubmed/23066108)
+
+#### Where to get help:
+
+- See the [mailing list](https://sourceforge.net/p/lofreq/mailman/)
 - See [the LoFreq blog](https://sourceforge.net/p/lofreq/blog/) for news and updates
-- non base-recalibrated data: can still run but expect FP,
-  especially with SB. We recommend GATK base-recalibration even for
-  non-human data or targeted sequencing. GATK usually requires the
-  input of known variant sites (circular, usually dbsnp) which
-  are not known for non-human data. one option: lofreq run (e.g. -B -J)
-  and use the resulting vcf as input.
-- quality of mapping
-- targeted sequencing, exome, bed files
-  If you want to run LoFreq only on certain regions use the
-  appropriate bed-file as input with -l. Not doing so will
-  negatively affect LoFreq sensitivity and it might calls variants
-  outside of the desired regions.
-- call-parallel fails with non-descript error message: run call
-  first and fix the command line. then execute call-parallel
-- If your data was heavily PCR-amplified LoFreq will likely call SNVs in the primer regions as well, due
-  to ambiguous primer positions, primer impurities etc. You should ignore primer positions after SNV
-  calling.
-- Another problem with heavily PCRed input is that PCR artifacts might show up as low-frequency SNVs,
-  especially if a mis-amplification happened in early cycles. Computational tools typically will be 
-  unable to distinguish these from true low-frequency variants, since the mis-incorporated bases look 
-  real to the sequencing machine.
+
+
+#### What happens if I didn't recalibrate base-qualities
+
+We usually recommend to follow GATK's best practices on post-processing of you BAM file which include base quality recalibration. You can run LoFreq on non-recalibrated data, which might however results in a higher rate of false positive calls. We would recommend GATK base-recalibration even for non-human data or targeted sequencing. GATK usually requires the input of known variant sites (which is a circular problem) which are however not known for most organisms non-human data (use dbsnp for human data). One option is to use initial LoFreq predictions as vcf input to GATK and then run LoFreq again.
+
+#### Regions for targeted sequencing
+
+If you want to run LoFreq only on certain regions use the appropriate bed-file as input with `-l regions.bed`. Not doing so will negatively affect LoFreq sensitivity and it might calls variants
+outside of the desired regions. 
+
+
+#### PCR amplified data
+
+If your data was heavily PCR-amplified LoFreq will likely call SNVs in the primer regions as well, due
+to ambiguous primer positions, primer impurities etc. You should ignore primer positions after SNV
+calling. The best way to achieve this is to create a bed-file that only lists non-primer regions and use that as input to Lofreq with `-l regions.bed`
+
+Another problem with heavily PCRed input is that PCR artifacts might show up as low-frequency SNVs,
+especially if a mis-amplification happened in early cycles. Computational tools will be 
+unable to distinguish these from true low-frequency variants, since the mis-incorporated bases look 
+real to the sequencing machine.
+
+### I get strange results with Bowtie. Why?
+
+Depending on the parameters you used for calling Bowtie it will sometimes produces alignments with reads that have either only mapping quality 0 (non unique) or MQ 255 (not available). It is unclear to us why that happens, or what exactly that means (see also [this thread](http://seqanswers.com/forums/showthread.php?t=3142)). Anyway, LoFreq will by default use mapping quality for the prediction of SNVs, which doesn't help in this scenario though. If you have to use Bowtie we suggest to switch off the use of mapping quality in LoFreq with `-J`
+
+
+#### `call-parallel` fails with a non-descript error message
+
+This is easiest to debug by running the single-threaded version `call` first with the same parameters, check the error message and fix the command accordingly. Then execute `call-parallel` again with fixed parameters.

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Tue, 01 Apr 2014 08:52:54 -0000

--- v3
+++ v4
@@ -1,7 +1,7 @@
 # UNDER CONSTRUCTION: FAQ for LoFreq*

 - how to cite
-- where to get help
+- where to get help: [https://sourceforge.net/p/lofreq/mailman/](mailing list)
 - See [the LoFreq blog](https://sourceforge.net/p/lofreq/blog/) for news and updates
 - non base-recalibrated data: can still run but expect FP,
   especially with SB. We recommend GATK base-recalibration even for

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Tue, 01 Apr 2014 08:24:59 -0000

--- v2
+++ v3
@@ -2,6 +2,7 @@

 - how to cite
 - where to get help
+- See [the LoFreq blog](https://sourceforge.net/p/lofreq/blog/) for news and updates
 - non base-recalibrated data: can still run but expect FP,
   especially with SB. We recommend GATK base-recalibration even for
   non-human data or targeted sequencing. GATK usually requires the

LoFreq-Star-FAQ modified by Andreas Wilm

Andreas Wilm — Tue, 01 Apr 2014 08:16:11 -0000