BMFtools

BMFtools (Barcoded Molecular Families tools) is a suite of tools for barcoded reads which takes advantage of PCR redundancy for error reduction/elimination. The core functionality consists of molecular demultiplexing at fastq stage, producing a unique observation for each sequenced founded template molecule. Accessory tools provide postprocessing, filtering, quality control, and summary statistics.

Installation

Requirements: gcc4.9+ samtools 1.2+

git clone https://github.com/ARUP-NGS/BMFtools --recursive
cd BMFtools
make

Use

bmftools

bmftools <--help/-h>

bmftools <subcommand> <-h>

Tools

Name	Use
bmftools cap	Postprocess a tagged BAM for BMF-agnostic tools.
bmftools depth	Calculates depth of coverage over a set of bed intervals.
bmftools collapse	Collapse initial fastq records by barcode
bmftools err	Calculate error rates based on cycle, base call, and quality score.
bmftools famstats	Calculate family size statistics for a bam alignment file.
bmftools filter	Filter or split a bam file by a set of filters.
bmftools mark	Add tags for rsq.
bmftools stack	A maximally-permissive variant caller using molecular barcode metadata analogous to samtools mpileup.
bmftools rsq	Rescue reads with using positional inference to collapse to unique observations in spite of errors in the barcode sequence.
bmftools sort	Sort for bam rescue
bmftools target	Calculates on-target rate.
bmftools vet	Curate variant calls from another variant caller (.bcf) and a bam alignment.

These tools are divided into four categories:

1. Core functionality
2. Manipulation
3. Analysis

Core Functionality

####bmftools collapse bmftools collapse combines reads sharing barcodes into single observations respectively.

First, the barcodes are added to the comment fields of the fastqs and split the records into subsets based on the first characters in the barcode. Then, reads with exactly-matching barcode are collapsed, with a meta-analysis performed on each base call.

bmftools collapse inline collapses templates where both strands were sequenced, whereas collapse secondary lacks strand information.

####bmftools rsq bmftools rsq uses positional information to collapse reads sharing alignment signatures with close barcodes under the assumption that they came from the same original founding molecule but with errors in reading the barcode.

Manipulation

####bmftools cap Caps quality scores using barcode metadata to facilitate working with barcode-agnostic tools.

####bmftools filter Filters or splits a bam file based on a set of filters. These can be inverted with -v (analogous to grep).

Filters:

• Fail reads with insufficient mapping quality.

• Fail reads with insufficient family size.

• Fail read pairs by aligned fraction.

• Fail reads outside of a bed region.

• Fail reads without all bits in given parameter in the sam flag field.

• Fail reads with any bits in given parameter in the sam flag field.

####bmftools vet Curates SNV calls from a tumor/normal variant call file using barcode metadata from the bams used to produce the variant call file.

Analysis

####bmftools depth Calculates depth of coverage across a bed file using barcode metadata.

####bmftools target Calculates on-target fraction for bed file using barcode metadata.

####bmftools err Calculates error rates by a variety of parameters. Additionally, pre-computes the quality score recalibration for the optional collapse recalibration step.

####bmftools famstats Calculates summary statistics related to family size and demultiplexing.

####bmftools stack A maximally-permissive variant caller using molecular barcode metadata analogous to samtools mpileup.

BMF Tags

Tag	Content	Format
DR	Whether the read was sequenced from both strands. Only valid for inline chemistry.	Integer [0, 1]
FA	Number of reads in Family which Agreed with final sequence at each base	uint32_t array
FM	Size of family (number of reads sharing barcode.), e.g., "Family Members"	Integer
FP	Read Passes Filter related to barcoding. Determines QC fail flag in bmftools mark (without -q).	Integer [0, 1]
NF	Mean number of differences between reads and consensus per read in family	Float
PV	Phred Values for a base call after meta-analysis	uint32_t array
RV	Number of reversed reads in consensus. Only for inline chemistry.	Integer