SNP calling exercise part 2 - Revision history

Mick at 16:20, 17 January 2026

2026-01-17T16:20:11Z

← Older revision		Revision as of 18:20, 17 January 2026
Line 168:		Line 168:
	<hr>		<hr>

	<p>Please find answers here:		<p>Please find answers here: [SNP_calling_exercise_part_2_answers SNP_calling_exercise_part_2_answers]</p>
	~~<a href="~~SNP_calling_exercise_part_2_answers">SNP_calling_exercise_part_2_answers~~</a>~~</p>

	<p><b>Congratulations — you finished the exercise!</b></p>		<p><b>Congratulations — you finished the exercise!</b></p>

Mick at 10:59, 7 January 2026

2026-01-07T10:59:25Z

← Older revision		Revision as of 12:59, 7 January 2026
Line 101:		Line 101:
	-a [INPUT VCF] \		-a [INPUT VCF] \
	-b /home/databases/databases/GRCh38/filter99.bed.gz \		-b /home/databases/databases/GRCh38/filter99.bed.gz \
	\| bgzip -c > [OUTPUT VCF]		\| /home/ctools/htslib-1.20/bgzip -c > [OUTPUT VCF]
	</pre>		</pre>

Line 127:		Line 127:
	GRCh38.99 \		GRCh38.99 \
	[INPUT VCF] \		[INPUT VCF] \
	\| bgzip -c > [OUTPUT VCF]		\| /home/ctools/htslib-1.20/bgzip -c > [OUTPUT VCF]
	</pre>		</pre>

Mick at 13:03, 6 January 2026

2026-01-06T13:03:45Z

← Older revision		Revision as of 15:03, 6 January 2026
Line 110:		Line 110:
	</pre>		</pre>

	<p>The "99" refers to the proportion of synthetic reads that map uniquely at that position~~. Explanation: [http://lh3lh3.users.sourceforge.net/snpable.shtml SNPable / mappability]~~.</p>		<p>The "99" refers to the proportion of synthetic reads that map uniquely at that position.</p>

	<h4>Q3</h4>		<h4>Q3</h4>

Mick at 10:16, 6 January 2026

2026-01-06T10:16:06Z

← Older revision		Revision as of 12:16, 6 January 2026
Line 152:		Line 152:

	<pre>		<pre>
	scp stud0XX@pupilX:path/to/NA24694_hf.html .		scp stud0XX@pupilX.healthtech.dtu.dk:path/to/NA24694_hf.html .
	</pre>		</pre>

Mick at 10:38, 5 January 2026

2026-01-05T10:38:31Z

← Older revision		Revision as of 12:38, 5 January 2026
Line 33:		Line 33:

	<pre>		<pre>
	gatk VariantFiltration \		/home/ctools/gatk-4.6.2.0/gatk VariantFiltration \
	-V [INPUT VCF] \		-V [INPUT VCF] \
	-O [OUTPUT VCF] \		-O [OUTPUT VCF] \

Mick at 14:33, 16 December 2025

2025-12-16T14:33:12Z

← Older revision		Revision as of 16:33, 16 December 2025
Line 135:		Line 135:
	</ul>		</ul>

	<p>Run snpEff on your hard-filtered VCF (before mappability filtering):</p>		<p>Run <code>snpEff</code> on your hard-filtered VCF (before mappability filtering).
			This produces:</p>

	<ul>		<ul>
Line 142:		Line 143:
	</ul>		</ul>

	<p>~~Open~~ the HTML report:</p>		<p><b>Viewing the snpEff HTML report:</b></p>

			<p>If you are using <b>MobaXterm</b>, you can open the HTML file directly from the
			left-hand file panel.</p>

			<p>If you are using <b>macOS</b> (or a standard terminal), copy the HTML file to
			your local computer and open it in a web browser. For example:</p>

	<pre>		<pre>
	~~firefox~~ NA24694_hf.html		scp stud0XX@pupilX:path/to/NA24694_hf.html .
	</pre>		</pre>

			<p>Replace <code>stud0XX</code> with your student ID and <code>pupilX</code> with the
			compute node you are working on.</p>


	<h4>Q4</h4>		<h4>Q4</h4>

Mick at 09:14, 26 November 2025

2025-11-26T09:14:19Z

Show changes

Mick at 09:12, 26 November 2025

2025-11-26T09:12:15Z

Show changes

Gabre at 15:27, 15 December 2024

2024-12-15T15:27:18Z

← Older revision		Revision as of 17:27, 15 December 2024
Line 4:		Line 4:
	We saw that the data contains some calls of poor quality. Ideally, we do not want to carry over these calls into downstream analyses. We will explore how to filter out genotype data.		We saw that the data contains some calls of poor quality. Ideally, we do not want to carry over these calls into downstream analyses. We will explore how to filter out genotype data.

	Please use the ~~vcf~~ file generated in part 1.		Please use the VCF file generated in part 1.


Line 61:		Line 61:
	In real life, there are no perfect filters, I suggest you progressively add them, measure their effectiveness and make sure that you do not introduce unwanted bias to your analyses.		In real life, there are no perfect filters, I suggest you progressively add them, measure their effectiveness and make sure that you do not introduce unwanted bias to your analyses.

	'''~~Q11~~'''		'''Q1'''

	How many sites have been filtered out? Remember that sites that passed the filter will have the string '''PASS''' as the seventh column, you can use '''grep''' with the trick we mentioned to remove lines with a specific string. How many SNPs were filtered out?		How many sites have been filtered out? Remember that sites that passed the filter will have the string '''PASS''' as the seventh column, you can use '''grep''' with the trick we mentioned to remove lines with a specific string. How many SNPs were filtered out?


	'''~~Q12~~'''		'''Q2'''

	The filtering command leaves in the seventh column a string (specified via --filter-name above) to determine which filter failed sites that were removed. Can you use the command above and modify it using the commands like '''cut''', '''sort''' and '''uniq''' to determine was the filter that filtered out the most sites? If you are starting to use UNIX, this question might be challenging.		The filtering command leaves in the seventh column a string (specified via --filter-name above) to determine which filter failed sites that were removed. Can you use the command above and modify it using the commands like '''cut''', '''sort''' and '''uniq''' to determine was the filter that filtered out the most sites? If you are starting to use UNIX, this question might be challenging.
Line 78:		Line 78:
	The "-header" just says print the header of [INPUT VCF]. Retain the sites that have '''passed the hard filtering''' that are contained in filter99.bed.gz and call the output NA24694_hf_map99.vcf.gz. The 99 is from the percentage of DNA fragments required to map uniquely at that particular position (read more [http://lh3lh3.users.sourceforge.net/snpable.shtml here].		The "-header" just says print the header of [INPUT VCF]. Retain the sites that have '''passed the hard filtering''' that are contained in filter99.bed.gz and call the output NA24694_hf_map99.vcf.gz. The 99 is from the percentage of DNA fragments required to map uniquely at that particular position (read more [http://lh3lh3.users.sourceforge.net/snpable.shtml here].

	'''~~Q13~~'''		'''Q3'''

	How many sites did you retain?		How many sites did you retain?
Line 104:		Line 104:
	and answer the following questions:		and answer the following questions:

	'''~~Q14~~'''		'''Q4'''

	What is the most common genomic region (exon, downstream, intron, UTR) of the variants we detected?		What is the most common genomic region (exon, downstream, intron, UTR) of the variants we detected?

	'''~~Q15~~'''		'''Q5'''

	How many variants can lead to a codon change? See explanation about point mutations on [https://en.wikipedia.org/wiki/Point_mutation Wikipedia]		How many variants can lead to a codon change? See explanation about point mutations on [https://en.wikipedia.org/wiki/Point_mutation Wikipedia]
Line 116:		Line 116:


	Please find answers [[~~SNP_calling_exercise_answers~~\|here]]		Please find answers [[SNP_calling_exercise_part_2_answers\|here]]

	'''Congratulations you finished the exercise!'''		'''Congratulations you finished the exercise!'''

Gabre at 15:25, 15 December 2024

2024-12-15T15:25:08Z

← Older revision		Revision as of 17:25, 15 December 2024
Line 4:		Line 4:
	We saw that the data contains some calls of poor quality. Ideally, we do not want to carry over these calls into downstream analyses. We will explore how to filter out genotype data.		We saw that the data contains some calls of poor quality. Ideally, we do not want to carry over these calls into downstream analyses. We will explore how to filter out genotype data.

			Please use the vcf file generated in part 1.


	<H3> Hard filtering </H3>		<H3> Hard filtering </H3>

	A major downside of soft filtering is that does not apply ~~for~~ samples for which we do not have a good representation of the genetic diversity. An alternative is to do hard filtering ~~meaning~~ filtering the variants according to predetermined cutoffs. This has the downside of potentially introducing bias if the filter is correlated with the type of variant (ex: if heterozygous sites have higher genotype quality).		We saw that soft filtering learns which variants are "true." However, a major downside of soft filtering is that it does not apply to samples for which we do not have a good representation of the genetic diversity.

			An alternative is to do hard filtering i.e. filtering the variants according to predetermined cutoffs. This has the downside of potentially introducing bias if the filter is correlated with the type of variant (ex: if heterozygous sites have higher genotype quality).

	First, we will consider the following mask file:		First, we will consider the following mask file: