Commit deda88a2 authored by Mathieu Giraud's avatar Mathieu Giraud

doc/vidjil-format.md: converted from doc/format-analysis.org + details

parent 2f952c41
#+TITLE: Encoding clones with V(D)J recombinations (2016b)
#+AUTHOR: The Vidjil team
#+HTML_HEAD: <link rel="stylesheet" type="text/css" href="org-mode.css" />
The following [[http://en.wikipedia.org/wiki/JSON][.json]] format allows to
encode a set of clones with V(D)J immune recombinations,
possibly with user annotations.
In Vidjil, this format is used by both the =.analysis= and the =.vidjil= files.
The =.vidjil= file represents the actual data on clones (and that can
reach megabytes, or even more), usually produced by processing reads by some RepSeq software.
(for example with detailed information on the 100 or 1000 top clones).
The =.analysis= file describes customizations done by the user
(or by some automatic pre-processing) on the Vidjil web application. The web application
can load or save such files (and possibly from/to the patient/sample database).
It is intended to be very small (a few kilobytes).
All settings in the =.analysis= file override the settings that could be
present in the =.vidjil= file.
* What is a clone ?
There are several definitions of what may be a clonotype,
depending on different RepSeq software or studies.
This format accept any kind of definition:
Clones are identified by a =id= string that may be an arbitrary identifier such as =clone-072a=.
Software computing clones may choose some relevant identifiers:
- =CGAGAGGTTACTATGATAGTAGTGGTTATTACGGGGTAGGGCAGTACTAC=, Vidjil algorithm, 50 nt window centered on the CDR3
- =CARPRDWNTYYYYGMDVW=, a CDR3 AA sequence
- =CARPRDWNTYYYYGMDVW IGHV3-11*00 IGHJ6*00=, a CDR3 AA sequence with additional V/J gene information (MiXCR)
- the 'clone sequence' as computed by the ARReST in =.clntab= files (processed by =fuse.py=)
- see also 'IMGT clonotype (AA) or (nt)'
* Examples
** =.vidjil= file -- one sample
This is an almost minimal =.vidjil= file, describing clones in one sample.
The =seg= element is optional: clones without =seg= elements will be shown on the grid with '?/?'.
The =_average_read_length= is also optional, but allows to plot GENSCAN-like plots more precisely than getting only the length of the sequence.
All other elements are required. The =reads.germlines= list can have only one element the case of data on a unique locus.
There is here one clone on the =TRG= locus with a designation =TRGV5*01 5/CC/0 TRGJ1*02=.
Note that other elements could be added by some program (such as =tag=, to identify some clones,
or =clusters=, to further cluster some clones, see below).
#+BEGIN_SRC js :tangle analysis-example1.vidjil
{
"producer": "program xyz version xyz",
"timestamp": "2014-10-01 12:00:11",
"vidjil_json_version": "2016b",
"samples": {
"number": 1,
"original_names": ["T8045-BC081-Diag.fastq"]
},
"reads" : {
"total" : [ 437164 ] ,
"segmented" : [ 335662 ] ,
"germline" : {
"TRG" : [ 250000 ] ,
"IGH" : [ 85662 ]
}
},
"clones": [
{
"id": "clone-001",
"sequence": "CTCATACACCCAGGAGGTGGAGCTGGATATTGATACTACGAAATCTAATTGAAAATGATTCTGGGGTCTATTACTGTGCCACCTGGGCCTTATTATAAGAAACTCTTTGGCAGTGGAAC",
"reads" : [ 243241 ],
"_average_read_length": [ 119.3 ],
"germline": "TRG",
"top": 1,
"seg":
{
"5": {"name": "TRGV5*01", "start": 1, "stop": 86, "delRight":5},
"3": {"name": "TRGJ1*02", "start": 89, "stop": 118, "delLeft":0},
"cdr3": { "start": 77, "stop": 104, "seq": "gccacctgggccttattataagaaactc" }
}
}
]
}
#+END_SRC
** =.vidjil= file -- several related samples
This a =.vidjil= file obtained by merging with =fuse.py= two =.vidjil= files corresponding to two samples.
Clones that have a same =id= are gathered (see 'What is a clone?', above).
It is the responsibility of the program generating the initial =.vidjil= files to choose these =id= to
do a correct gathering.
#+BEGIN_SRC js :tangle analysis-example2.vidjil
{
"producer": "program xyz version xyz / fuse.py version xyz",
"timestamp": "2014-10-01 14:00:11",
"vidjil_json_version": "2016b",
"samples": {
"number": 2,
"original_names": ["T8045-BC081-Diag.fastq", "T8045-BC082-fu1.fastq"]
},
"reads" : {
"total" : [ 437164, 457810 ] ,
"segmented" : [ 335662, 410124 ] ,
"germline" : {
"TRG" : [ 250000, 300000 ] ,
"IGH" : [ 85662, 10124 ]
}
},
"clones": [
{
"id": "clone-001",
"sequence": "CTCATACACCCAGGAGGTGGAGCTGGATATTGATACTACGAAATCTAATTGAAAATGATTCTGGGGTCTATTACTGTGCCACCTGGGCCTTATTATAAGAAACTCTTTGGCAGTGGAAC",
"reads" : [ 243241, 14717 ],
"germline": "TRG",
"top": 1,
"seg":
{
"5": {"name": "TRGV5*01", "start": 1, "stop": 86, "delRight": 5},
"3": {"name": "TRGJ1*02", "start": 89, "stop": 118, "delLeft": 0}
}
},
{
"id": "clone2",
"sequence": "GATACA",
"reads": [ 153, 10221 ],
"germline": "TRG",
"top": 2
},
{
"id": "clone3",
"sequence": "ATACAGA",
"reads": [ 521, 42 ],
"germline": "TRG",
"top": 3,
"seg":
{
"5": {"start": 1, "stop": 100},
"3": {"start": 101, "stop": 200}
}
}
]
}
#+END_SRC
** =.analysis= file
This file reflects the annotations a user could have done within the Vidjil web application or some other tool.
She has manually set sample names (=names=), tagged (=tag=, =tags=), named (=name=) and clustered (=clusters=)
some clones, and added external data (=data=).
#+BEGIN_SRC js :tangle analysis-example2.analysis
{
"producer": "user Bob, via Vidjil webapp",
"timestamp": "2014-10-01 12:00:11",
"vidjil_json_version": "2016b",
"samples": {
"id": [
"T8045-BC081-Diag.fastq",
"T8045-BC082-fu1.fastq"
],
"number": 2,
"names": ["diag", "fu1"],
"original_names": ["file1.fastq", "file2.fastq"],
"order": [1, 0]
},
"clones": [
{
"id": "clone-001",
"name": "Main ALL clone",
"tag": "0"
},
{
"id": "spikeE",
"label": "spike",
"sequence": "ATGACTCTGGAGTCTATTACTGTGCCACCTGGGATGTGAGTATTATAAGAAAC",
"tag": "3",
"expected": "0.1"
}
],
"clusters": [
[ "clone2", "clone3"],
[ "clone-5", "clone-10", "clone-179" ]
],
"data": {
"qPCR": [0.83, 0.024],
"spikeZ": [0.01, 0.02]
},
"tags": {
"names": {
"0" : "main clone",
"3" : "spike",
"5" : "custom tag"
},
"hide": [4, 5]
}
}
#+END_SRC
The =order= field defines the order in which order the points should be
considered. In that case we should first consider the second point (whose =name=
is /fu1)/ and the point to be considered in second should be the first one in
the file (whose =name= is /diag/).
The =clusters= field indicate clones (by their =id=) that have been further clustered.
Usually, these clones were defined in a related =.vidjil= file (as /clone2/ and /clone3/,
see the =.vidjil= file in the previous section). If these clones do not exist, the clusters are
just ignored. The first item of the cluster is considered as the
representative clone of the cluster.
* Detailed specification
** Generic information for traceability [required]
#+BEGIN_SRC js
"producer": "my-repseq-software -z -k (v. 123)", // arbitrary string, user/software/version/options producing this file [required]
"timestamp": "2014-10-01 12:00:11", // last modification date [required]
"vidjil_json_version": "2016b", // version of the .json format [required]
#+END_SRC
** Statistics: the =reads= element [.vidjil only, required]
The number of analyzed reads (=segmented=) may be higher than the sum of the read number of all clones,
when one choose to report only the 'top' clones (=-t= option for fuse).
#+BEGIN_SRC js
{
"total" : [], // total number of reads per sample (with samples.number elements)
"segmented" : [], // number of analyzed/segmented reads per sample (with samples.number elements)
"germline" : { // number of analyzed/segmented reads per sample/germline (with samples.number elements)
"TRG" : [],
"IGH" : []
}
}
#+END_SRC
** =samples= element [required]
#+BEGIN_SRC js
{
"number": 2, // number of samples [required]
"original_names": [], // original sample names (with samples.number elements) [required]
"names": [], // custom sample names (with samples.number elements) [optional]
// These names are editable and will be used on the graphs
"order": [], // custom sample order (lexicographic order by default) [optional]
// traceability on each sample (with sample.number elements)
"producer": [],
"timestamp": [],
"log": []
}
#+END_SRC
** =clones= list, with read count, tags, V(D)J designation and other sequence features
Each element in the =clones= list describes properties of a clone.
In a =.vidjil= file, this is the main part, describing all clones.
In the =.analysis= file, this section is intended to describe some specific clones.
#+BEGIN_SRC js
{
"id": "", // clone identifier, must be unique [required] [see above, 'What is a clone ?']
// the clone identifier in the .vidjil file and in .analysis file must match
"germline": "" // [required for .vidjil]
// (should match a germline defined in germline/germline.data)
"name": "", // clone custom name [optional]
// (the default name, in .vidjil, is computed from V/D/J information)
"label": "", // clone labels, separed by spaces [optional]
// These labels may add some information entered with a controled vocabulary
"sequence": "", // reference nt sequence [required for .vidjil]
// (for .analysis, not really used now in the web application,
// for special clones/sequences that are known,
// such as standard/spikes or know patient clones)
"tag": "", // tag id from 0 to 7 (see below) [optional]
"expected": "" // expected abundance of this clone (between 0 and 1) [optional]
// this will create a normalization option in the
// settings web application menu
"seg": // detailed V(D)J designation/segmentation and other sequences features or values [optional]
// on the web application, clones that are not segmented will be shown on the grid with '?/?'
// positions are related to the 'sequence'
// names of V/D/J genes should match the ones in files referenced in germline/germline.data
// Positions on the sequence start at 1.
{
"5": {"name": "IGHV5*01", "start": 1, "stop": 120, "delRight": 5}, // V (or 5') segment
"4": {"name": "IGHD1*01", "start": 124, "stop": 135, "info": "unsure designation", "delRight": 5, "delLeft": 0}, // D (or middle) segment
// Recombination with several D may use "4a", "4b"...
"3": {"name": "IGHJ3*02", "start": 136, "stop": 171, "delLeft": 5}, // J (or 3') segment
// any feature to be highlighted in the sequence, with optional fields related to this feature:
// - "start"/"stop" : positions on the clone sequence (starting at 1)
// - "delLeft/delRight" : a numerical value . It is the numbers of nucleotides deleted during the rearrangment. DelRight are compatible with V/5 and D/4 segments, delLeft is compatible with D/4 and J/3 segments.
// - "seq" : a sequence
// - "val" : a numerical value
// - "info" : a textual vlaue
//
// JUNCTION//CDR3 should be stored that way (in fields called "junction" of "cdr3"),
// its productivity must be stored in a boolean field called "productive".
"somefeature": { "start": 56, "stop": 61, "seq": "ACTGTA", "val": 145.7, "info": "analyzed with xyz" },
// Numerical or textual features concerning all the sequence or its analysis (such as 'evalue')
// can be provided by omitting "start" and "stop" elements.
"someotherfeature": {"val": 0.004521},
"anotherfeature": {"info": "VH CDR3 stereotypy"},
}
"reads": [], // number of reads in this clones [.vidjil only, required]
// (with samples.number elements)
"top": 0, // (not documented now) [required] threshold to display/hide the clone
"stats": [] // (not documented now) [.vidjil only] (with sample.number elements)
}
#+END_SRC
** =germlines= list [optional][work in progress, to be documented]
extend the =germline.data= default file with a custom germline
#+BEGIN_SRC js
"germlines" : {
"custom" : {
"shortcut": "B",
"5": ["TRBV.fa"],
"4": ["TRBD.fa"],
"3": ["TRBJ.fa"]
}
}
#+END_SRC
** Further clustering of clones: the =clusters= list [optional]
Each element in the 'clusters' list describe a list of clones that are 'merged'.
In the web application, it will be still possible to see them or to unmerge them.
The first clone of each line is used as a representative for the cluster.
** =data= list [optional][work in progress, to be documented]
Each element in the =data= list is a list of values (of size samples.number)
showing additional data for each sample, as for example qPCR levels or spike information.
In the browser, it will be possible to display these data and to normalize
against them (not implemented now).
** Tagging some clones: =tags= list [optional]
The =tags= list describe the custom tag names as well as tags that should be hidden by default.
The default tag names are defined in [[../browser/js/vidjil-style.js]].
#+BEGIN_SRC js
"key" : "value" // "key" is the tag id from 0 to 7 and "value" is the custom tag name attributed
#+END_SRC
......@@ -491,7 +491,7 @@ two windows that must be clustered.
The main output of Vidjil-algo (with the default `-c clones` command) are two following files:
- The `.vidjil` file is *the file for the Vidjil web application*.
The file is in a `.json` format (detailed in [format-analysis.org](format-analysis.org))
The file is in a `.json` format (detailed in [vidjil-format](vidjil-format))
describing the windows and their count, the consensus sequences (`-y`),
the detailed V(D)J and CDR3 designation (`-z`, see warning below), and possibly
the results of the further clustering.
......
# Encoding clones with V(D)J recombinations with `.vidjil` files
The following [.json](http://en.wikipedia.org/wiki/JSON) format (2016b) allows to
encode a set of clones with V(D)J immune recombinations,
possibly with user annotations.
In Vidjil, this format is used by both the `.analysis` and the `.vidjil` files.
The `.vidjil` file represents the actual data on clones (and that can
reach megabytes, or even more), usually produced by processing reads by some RepSeq software.
(for example with detailed information on the 100 or 1000 top clones).
The `.analysis` file describes customizations done by the user
(or by some automatic pre-processing) on the Vidjil web application. The web application
can load or save such files (and possibly from/to the patient/sample database).
It is intended to be very small (a few kilobytes).
All settings in the `.analysis` file override the settings that could be
present in the `.vidjil` file.
# What is a clone ?
There are several definitions of what may be a clonotype,
depending on different RepSeq software or studies.
This format accept any kind of definition:
Clones are identified by a `id` string that may be an arbitrary identifier such as `clone-072a`.
Software computing clones may choose some relevant identifiers:
- `CGAGAGGTTACTATGATAGTAGTGGTTATTACGGGGTAGGGCAGTACTAC`, Vidjil algorithm, 50 nt window centered on the CDR3
- `CARPRDWNTYYYYGMDVW`, a CDR3 AA sequence
- `CARPRDWNTYYYYGMDVW IGHV3-11*00 IGHJ6*00`, a CDR3 AA sequence with additional V/J gene information (MiXCR)
- the 'clone sequence' as computed by the ARReST in `.clntab` files (processed by `fuse.py`)
- see also 'IMGT clonotype (AA) or (nt)'
# Examples
## `.vidjil` file – one sample
This is an almost minimal `.vidjil` file, describing clones in one sample.
The `seg` element is optional: clones without `seg` elements will be shown on the grid with '?/?'.
The `_average_read_length` is also optional, but allows to plot GENSCAN-like plots more precisely than getting only the length of the sequence.
All other elements are required. The `reads.germlines` list can have only one element the case of data on a unique locus.
There is here one clone on the `TRG` locus with a designation `TRGV5*01 5/CC/0 TRGJ1*02`.
Note that other elements could be added by some program (such as `tag`, to identify some clones,
or `clusters`, to further cluster some clones, see below).
``` javascript
{
"producer": "program xyz version xyz",
"timestamp": "2014-10-01 12:00:11",
"vidjil_json_version": "2016b",
"samples": {
"number": 1,
"original_names": ["T8045-BC081-Diag.fastq"]
},
"reads" : {
"total" : [ 437164 ] ,
"segmented" : [ 335662 ] ,
"germline" : {
"TRG" : [ 250000 ] ,
"IGH" : [ 85662 ]
}
},
"clones": [
{
"id": "clone-001",
"sequence": "CTCATACACCCAGGAGGTGGAGCTGGATATTGATACTACGAAATCTAATTGAAAATGATTCTGGGGTCTATTACTGTGCCACCTGGGCCTTATTATAAGAAACTCTTTGGCAGTGGAAC",
"reads" : [ 243241 ],
"_average_read_length": [ 119.3 ],
"germline": "TRG",
"top": 1,
"seg":
{
"5": {"name": "TRGV5*01", "start": 1, "stop": 86, "delRight":5},
"3": {"name": "TRGJ1*02", "start": 89, "stop": 118, "delLeft":0},
"cdr3": { "start": 77, "stop": 104, "seq": "gccacctgggccttattataagaaactc" }
}
}
]
}
```
## `.vidjil` file – several related samples
This a `.vidjil` file obtained by merging with `fuse.py` two `.vidjil` files corresponding to two samples.
Clones that have a same `id` are gathered (see 'What is a clone?', above).
It is the responsibility of the program generating the initial `.vidjil` files to choose these `id` to
do a correct gathering.
``` javascript
{
"producer": "program xyz version xyz / fuse.py version xyz",
"timestamp": "2014-10-01 14:00:11",
"vidjil_json_version": "2016b",
"samples": {
"number": 2,
"original_names": ["T8045-BC081-Diag.fastq", "T8045-BC082-fu1.fastq"]
},
"reads" : {
"total" : [ 437164, 457810 ] ,
"segmented" : [ 335662, 410124 ] ,
"germline" : {
"TRG" : [ 250000, 300000 ] ,
"IGH" : [ 85662, 10124 ]
}
},
"clones": [
{
"id": "clone-001",
"sequence": "CTCATACACCCAGGAGGTGGAGCTGGATATTGATACTACGAAATCTAATTGAAAATGATTCTGGGGTCTATTACTGTGCCACCTGGGCCTTATTATAAGAAACTCTTTGGCAGTGGAAC",
"reads" : [ 243241, 14717 ],
"germline": "TRG",
"top": 1,
"seg":
{
"5": {"name": "TRGV5*01", "start": 1, "stop": 86, "delRight": 5},
"3": {"name": "TRGJ1*02", "start": 89, "stop": 118, "delLeft": 0}
}
},
{
"id": "clone2",
"sequence": "GATACA",
"reads": [ 153, 10221 ],
"germline": "TRG",
"top": 2
},
{
"id": "clone3",
"sequence": "ATACAGA",
"reads": [ 521, 42 ],
"germline": "TRG",
"top": 3,
"seg":
{
"5": {"start": 1, "stop": 100},
"3": {"start": 101, "stop": 200}
}
}
]
}
```
## `.analysis` file
This file reflects the annotations a user could have done within the Vidjil web application or some other tool.
She has manually set sample names (`names`), tagged (`tag`, `tags`), named (`name`) and clustered (`clusters`)
some clones, and added external data (`data`).
``` javascript
{
"producer": "user Bob, via Vidjil webapp",
"timestamp": "2014-10-01 12:00:11",
"vidjil_json_version": "2016b",
"samples": {
"id": [
"T8045-BC081-Diag.fastq",
"T8045-BC082-fu1.fastq"
],
"number": 2,
"names": ["diag", "fu1"],
"original_names": ["file1.fastq", "file2.fastq"],
"order": [1, 0]
},
"clones": [
{
"id": "clone-001",
"name": "Main ALL clone",
"tag": "0"
},
{
"id": "spikeE",
"label": "spike",
"sequence": "ATGACTCTGGAGTCTATTACTGTGCCACCTGGGATGTGAGTATTATAAGAAAC",
"tag": "3",
"expected": "0.1"
}
],
"clusters": [
[ "clone2", "clone3"],
[ "clone-5", "clone-10", "clone-179" ]
],
"data": {
"qPCR": [0.83, 0.024],
"spikeZ": [0.01, 0.02]
},
"tags": {
"names": {
"0" : "main clone",
"3" : "spike",
"5" : "custom tag"
},
"hide": [4, 5]
}
}
```
The `order` field defines the order in which order the points should be
considered. In that case we should first consider the second point (whose `name`
is *fu1)* and the point to be considered in second should be the first one in
the file (whose `name` is *diag*).
The `clusters` field indicate clones (by their `id`) that have been further clustered.
Usually, these clones were defined in a related `.vidjil` file (as *clone2* and *clone3*,
see the `.vidjil` file in the previous section). If these clones do not exist, the clusters are
just ignored. The first item of the cluster is considered as the
representative clone of the cluster.
# Detailed specification
## Generic information for traceability \[required\]
``` javascript
"producer": "my-repseq-software -z -k (v. 123)", // arbitrary string, user/software/version/options producing this file [required]
"timestamp": "2014-10-01 12:00:11", // last modification date [required]
"vidjil_json_version": "2016b", // version of the .json format [required]
```
## Statistics: the `reads` element \[.vidjil only, required\]
The number of analyzed reads (`segmented`) may be higher than the sum of the read number of all clones,
when one choose to report only the 'top' clones (`-t` option for fuse).
``` javascript
{
"total" : [], // total number of reads per sample (with samples.number elements)
"segmented" : [], // number of analyzed/segmented reads per sample (with samples.number elements)
"germline" : { // number of analyzed/segmented reads per sample/germline (with samples.number elements)
"TRG" : [],
"IGH" : []
}
}
```
## `samples` element \[required\]
``` javascript
{
"number": 2, // number of samples [required]
"original_names": [], // original sample names (with samples.number elements) [required]
"names": [], // custom sample names (with samples.number elements) [optional]
// These names are editable and will be used on the graphs
"order": [], // custom sample order (lexicographic order by default) [optional]
// traceability on each sample (with sample.number elements)
"producer": [],
"timestamp": [],
"log": []
}
```
## `clones` list, with read count, tags, V(D)J designation and other sequence features
Each element in the `clones` list describes properties of a clone.