`host` parameter ignored by igvR constructor
Closed this issue · 16 comments
The host parameter specified by the igvR constructor appears to be ignored and this is leading to errors displaying tracks:
igv <- igvR(host = "192.168.0.1", quiet=TRUE)
igv@uri
# "http://localhost:15002"
# ... setup genome and tracks
displayTrack(igv, my_bam_track)
# bam url: http://localhost:15003?/tmp/Rtmp0ZcCIP/filec49620e6c76a.bam
# bam track height: 50
Note that the uri slot and the displayTrack message both say "localhost" instead of the host value I specified. The actual viewer also displays an error:
Error accessing resource: http://localhost:15003?/tmp/Rtmp0ZcCIP/filec49620e6c76a.bam.bai Status: 0
sessionInfo() is:
R version 4.0.2 (2020-06-22)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 18.04.5 LTS
Matrix products: default
BLAS: /usr/lib/x86_64-linux-gnu/openblas/libblas.so.3
LAPACK: /usr/lib/x86_64-linux-gnu/libopenblasp-r0.2.20.so
locale:
[1] LC_CTYPE=C.UTF-8 LC_NUMERIC=C LC_TIME=C.UTF-8 LC_COLLATE=C.UTF-8 LC_MONETARY=C.UTF-8
[6] LC_MESSAGES=C.UTF-8 LC_PAPER=C.UTF-8 LC_NAME=C LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=C.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] parallel stats4 stats graphics grDevices utils datasets methods base
other attached packages:
[1] BSgenome.Hsapiens.UCSC.hg38_1.4.3 BSgenome_1.58.0 rtracklayer_1.50.0
[4] igvR_1.15.3 BrowserViz_2.12.0 httpuv_1.6.5
[7] jsonlite_1.7.3 GenomicAlignments_1.26.0 SummarizedExperiment_1.20.0
[10] Biobase_2.50.0 MatrixGenerics_1.2.1 matrixStats_0.61.0
[13] reticulate_1.16 data.table_1.14.2 forcats_0.5.0
[16] stringr_1.4.0 dplyr_1.0.7 purrr_0.3.4
[19] readr_1.3.1 tidyr_1.1.1 tibble_3.1.6
[22] ggplot2_3.3.2 tidyverse_1.3.0 Rsamtools_2.6.0
[25] Biostrings_2.58.0 XVector_0.30.0 GenomicRanges_1.42.0
[28] GenomeInfoDb_1.26.7 IRanges_2.24.1 S4Vectors_0.28.1
[31] BiocGenerics_0.36.1
loaded via a namespace (and not attached):
[1] colorspace_2.0-2 ellipsis_0.3.2 fs_1.5.2 rstudioapi_0.13 farver_2.1.0
[6] bit64_4.0.5 AnnotationDbi_1.52.0 fansi_1.0.2 lubridate_1.7.9 xml2_1.3.3
[11] cachem_1.0.6 knitr_1.29 splitstackshape_1.4.8 seqLogo_1.56.0 broom_0.7.0
[16] dbplyr_2.1.1 compiler_4.0.2 httr_1.4.2 backports_1.4.1 assertthat_0.2.1
[21] Matrix_1.4-0 fastmap_1.1.0 cli_3.1.1 later_1.3.0 prettyunits_1.1.1
[26] tools_4.0.2 gtable_0.3.0 glue_1.6.1 GenomeInfoDbData_1.2.4 rappdirs_0.3.3
[31] Rcpp_1.0.8 cellranger_1.1.0 vctrs_0.3.8 xfun_0.16 rvest_0.3.6
[36] lifecycle_1.0.1 XML_3.99-0.8 zlibbioc_1.36.0 scales_1.1.1 VariantAnnotation_1.36.0
[41] hms_1.1.1 promises_1.2.0.1 RColorBrewer_1.1-2 curl_4.3.2 memoise_2.0.1
[46] biomaRt_2.46.3 stringi_1.7.6 RSQLite_2.2.9 GenomicFeatures_1.42.3 BiocParallel_1.24.1
[51] rlang_1.0.0 pkgconfig_2.0.3 bitops_1.0-7 lattice_0.20-45 labeling_0.4.2
[56] bit_4.0.4 tidyselect_1.1.1 magrittr_2.0.2 R6_2.5.1 generics_0.1.1
[61] DelayedArray_0.16.3 DBI_1.1.2 pillar_1.6.5 haven_2.3.1 withr_2.4.3
[66] RCurl_1.98-1.5 MotifDb_1.32.0 modelr_0.1.8 crayon_1.4.2 utf8_1.2.2
[71] BiocFileCache_1.14.0 progress_1.2.2 grid_4.0.2 readxl_1.3.1 blob_1.2.2
[76] reprex_0.3.0 digest_0.6.29 openssl_1.4.6 munsell_0.5.0 viridisLite_0.3.0
[81] askpass_1.1
EDIT:
I manually edited the contents of the uri slot, but am still getting the viewer error. The path in the URL "/tmp/Rtmp0ZcCIP/filec49620e6c76a.bam" is the correct file. When I use lsof on the command line, I notice that rsession is still listening on localhost instead of the ip, so I think that is still the issue.
@genec1 - looking at this now, thanks for the report.
@genec1,
The constructor's host argument is a bad idea - my bad idea, left over from a long time ago. The man page on the constructor illuminates this, partly:
@param host In practice, this is always "localhost"
The justification: igvR displays in your browser, and it is hard for me to imagine a case when "your browser" runs on anything but localhost. Do you have a use-case that suggests otherwise? I'd be glad to see that, and to see if I can support it.
Even with this historical relic cleaned up, I have not helped you with your problem. Let me make sure I understand what that is. The best way would be for you to include the code with which you construct your GenomicAlignmentsTrack object. Here are two examples from inst/unitTests/test_igvR.R.
First, display an alignment already read into memory:
little.region <- GRanges(seqnames="21", ranges=IRanges(10400126, 10400326))
param <- ScanBamParam(which=little.region, what=scanBamWhat())
x <- readGAlignments(bamFile, use.names=TRUE, param=param)
track <- GenomicAlignmentTrack("bam demo", x, visibilityWindow=2000000, trackHeight=500) # 30000 default
displayTrack(igv, track)
Second, display an alignment hosted elsewhere:
url <- "https://1000genomes.s3.amazonaws.com/phase3/data/HG02450/alignment/HG02450.mapped.ILLUMINA.bwa.ACB.low_coverage.20120522.bam"
index <- "https://1000genomes.s3.amazonaws.com/phase3/data/HG02450/alignment/HG02450.mapped.ILLUMINA.bwa.ACB.low_coverage.20120522.bam.bai"
track <- RemoteAlignmentTrack("remote bam", url, index, visibilityWindow=100000)
displayTrack(igv, track)
How do you create your track object?
@genec1, The constructor's host argument is a bad idea - my bad idea, left over from a long time ago. The man page on the constructor illuminates this, partly:
@param host In practice, this is always "localhost"The justification: igvR displays in your browser, and it is hard for me to imagine a case when "your browser" runs on anything but localhost. Do you have a use-case that suggests otherwise? I'd be glad to see that, and to see if I can support it.
My particular use case is I’m running the code on an instance on AWS, as my local machine can’t handle the amount of data I’m working with.
Even with this historical relic cleaned up, I have not helped you with your problem. Let me make sure I understand what that is. The best way would be for you to include the code with which you construct your GenomicAlignmentsTrack object. Here are two examples from
inst/unitTests/test_igvR.R.
How do you create your track object?
I’m working with local bam files read into GenomicAlignment objects very similar to your first example. I will share actual code when I’m back on my workstation.
Thanks
Bam track display code:
# libraries
library(igvR)
library(BSgenome.Hsapiens.UCSC.hg38)
# setup igv
igv <- igvR(quiet=TRUE)
setGenome(igv, "hg38")
showGenomicRegion(igv, "chr17:7,663,654-7,692,258")
loc <- getGenomicRegion(igv)
logGR <- with(loc, GRanges(seqnames=chrom, ranges = IRanges(start, end)))
param <- ScanBamParam(which=logGR, what = scanBamWhat())
# setup bam
# <define BAM_FILE_PATH and BAM_TRACK_NAME>
galign <- readGAlignments(BAM_FILE_PATH, use.names=TRUE, param=param)
track <- GenomicAlignmentTrack(BAM_TRACK_NAME, galign)
displayTrack(igv, track)
# bam url: http://localhost:15001?/tmp/Rtmpqy9pbo/fileae83dff0a45.bam
# bam track height: 50
# viewer message: Error accessing resource: http://localhost:15001?/tmp/Rtmpqy9pbo/fileae83dff0a45.bam.bai Status: 0
BAM_FILE_PATH is the full path to a local file on the AWS instance where I'm running R, so I don't think it will be of help to you.
No problem. I'm working with multiple files, but here is one:
/home/rstudio/s3/HEK293T/SRR844_1.filtered.aligned.unique.bam
1 + 2 + 3a
Specifically, I am coding in a local web browser running RStudio, backed by rstudio-server on the AWS instance. When I initiate igvR in my R session, a new window opens in my local web browser showing IGV. This new window correctly shows the reference genome, but cannot show the programmatically loaded bam data since that is not on localhost relative to my browser session.
As a stopgap, I have been region-filtering the bams on AWS, sftp'ing them to my local machine, and opening them in a local IGV app. This is cumbersome for multiple bams with multiple regions. If you can point me to the byte-range/CORS server info, that would be appreciated!
@genec1
I have a presentation to give tomorrow, so today I can only give you quick & dirty instructions on creating a python flask webserver for serving up indexed genomic files, bam and vcf, for instance. I will polish these up by the end of the week and add them to this repo's README. I hope the following gets you going. Let me know:
- python >= version 3
- two extra modules: Flask and flask_cors
- your webserver app, which I call
serveStaticGenomeFiles.py:
from flask import Flask
from flask_cors import CORS
app = Flask(__name__, static_url_path='/static')
CORS(app)
@app.route('/')
def serveStaticFiles():
return 'CORS and byte-range request flask webserver for igvR\n'
if __name__ == "__main__":
app.run(host='0.0.0.0', port='60050')
- adjust the port to something appropriate
- run the python script with your files in
/staticof whatever path you specify when you construct the ```app``
I run all of this out of Docker container, and will document that as well.
I hope this quick description is helpful. I will incorporate your judgement, and appropriate improvements, when I have a chance to do this later in the week.
This is very much appreciated! I do not have the time to try it at the moment, but will give it a go in a week or two.
@genec1 I think I have figured out a few changes needed to run igvR from a remote RStudio (or plain R) session. I will be testing in the next few days. I hope this will be useful to you.
@genec1
igvR 1.15.8 and BrowserViz 2.17.1 just passed all tests for me, running in a remote RStudio Server session.
Give it a try?
@paul-shannon I can confirm that it works with the GitHub versions when using the host parameter and running on a remote host. Awesome!