afafa0301ea4b14fbe1fbd5aa379c5351ecd640d angie Tue Aug 2 19:41:44 2022 -0700 Add support for non-wuhCor1 genomes (e.g. monkeypox GenArk hub). * Search in hgPhyloPlaceData for config.ra files, taking assembly name (minus hub prefix) from directory name. * Add a menu input to the main page for switching between supported genomes if there are more than one. * Replace hardcoded values or global vars with dnaSeq attributes, assembly metadata queries or new config.ra settings. * Separate out SARS-CoV-2-specific help text like GISAID/CNCB descriptions. * Support metadata columns for GenBank-specific stuff & Nextstrain lineages (for MPXV). * also a little refactoring in runUsher in preparation for supporting usher server mode: parse new placement info file so we don't have to parse that data form usher stderr output. TODO: update Nextstrain/Auspice JSON output to use appropriate metadata columns and support monkeypox genes. diff --git src/hg/hgPhyloPlace/hgPhyloPlace.c src/hg/hgPhyloPlace/hgPhyloPlace.c index 2dc6bba..64b1e8e 100644 --- src/hg/hgPhyloPlace/hgPhyloPlace.c +++ src/hg/hgPhyloPlace/hgPhyloPlace.c @@ -1,16 +1,16 @@ -/* hgPhyloPlace - Upload SARS-CoV-2 sequence for placement in phylo tree. */ +/* hgPhyloPlace - Upload SARS-CoV-2 or MPXV sequence for placement in phylo tree. */ /* Copyright (C) 2020 The Regents of the University of California */ #include "common.h" #include "botDelay.h" #include "cart.h" #include "cgiApoptosis.h" #include "cheapcgi.h" #include "hCommon.h" #include "hash.h" #include "hui.h" #include "jsHelper.h" #include "knetUdc.h" #include "linefile.h" #include "net.h" @@ -59,30 +59,78 @@ else if (isNotEmpty(fileBinaryCoords)) { fprintf(stderr, "%s=%s fileBinaryCoords=%s\n", cartVar, fileName, fileBinaryCoords); char *binInfo = cloneString(fileBinaryCoords); char *words[2]; char *mem; unsigned long size; chopByWhite(binInfo, words, ArraySize(words)); mem = (char *)sqlUnsignedLong(words[0]); size = sqlUnsignedLong(words[1]); lf = lineFileDecompressMem(TRUE, mem, size); } return lf; } +static char *labelForDb(char *db) +/* The assembly hub name is just the accession; make a special label for hMPXV. Otherwise just + * return hGenome(db). */ +{ +char *label = NULL; +if (sameString(trackHubSkipHubName(db), "GCF_014621545.1")) + label = cloneString("hMPXV"); +else + label = hGenome(db); +return label; +} + +static void selectDb(char **pDb, char **pLabel) +/* Search for assembly config.ra files in hgPhyloPlaceData. If there is more than one + * supported assembly, then make a menu / select input for supported assemblies; + * reload the page on change. */ +{ +struct slName *supportedDbs = phyloPlaceDbList(); +if (supportedDbs == NULL) + errAbort("Sorry, this server is not configured to perform phylogenetic placement."); +if (!slNameInList(supportedDbs, *pDb)) + { + *pDb = cloneString(supportedDbs->name); + } +*pLabel = labelForDb(*pDb); +int supportedDbCount = slCount(supportedDbs); +if (supportedDbCount > 1) + { + char *labels[supportedDbCount]; + char *values[supportedDbCount]; + struct slName *sDb; + int i; + for (sDb = supportedDbs, i = 0; i < supportedDbCount; sDb = sDb->next, i++) + { + values[i] = sDb->name; + labels[i] = labelForDb(values[i]); + } + char *selectVar = "db"; + struct dyString *dy = jsOnChangeStart(); + jsDropDownCarryOver(dy, selectVar); + char *js = jsOnChangeEnd(&dy); + puts("<p>Choose your pathogen: "); + cgiMakeDropListFull(selectVar, labels, values, supportedDbCount, *pDb, "change", js); + puts("</p>"); + } +slNameFreeList(&supportedDbs); +} + static void newPageStartStuff() { // Copied these from hgGtexTrackSettings.c which says "// NOTE: This will likely go to web.c". puts("<link rel='stylesheet' href='../style/gb.css'>"); puts("<link rel='stylesheet' href='../style/hgGtexTrackSettings.css'>"); //#*** TODO: move this out to a CSS (hardcoding for now because we're doing a standalone push //#*** independent of the release cycle). puts("<style>\n" "#warnBox {\n" " border: 3px ridge DarkRed;\n" " width:640px;\n" " padding:10px; \n" " margin:10px;\n" " text-align:left;\n" @@ -171,178 +219,208 @@ "{ var $fileInput = $('input[name="fileVarName"]');" \ " var $pasteInput = $('textarea[name="pasteVarName"]');" \ " if ($fileInput && $fileInput[0] && $fileInput[0].files && !$fileInput[0].files.length &&" \ " $pasteInput && !$pasteInput.val()) {" \ " alert('Please either choose a file or paste in sequence names/IDs first, ' +" \ " 'and then click the upload button.');" \ " return false; " \ " } else if ($fileInput && $fileInput[0] && $fileInput[0].files && " \ " !!$fileInput[0].files.length &&" \ " $pasteInput && !!$pasteInput.val()) {" \ " alert('Sorry, unable to process both a file and pasted-in sequence names/IDs at the ' +" \ " 'same time. Please clear one or the other and then click the upload button.');" \ " return false; " \ " } else { loadingImage.run(); return true; } }" -static void inputForm() +static void inputForm(char *db) /* Ask the user for FASTA or VCF. */ { printf("<form action='%s' name='mainForm' method=POST enctype='multipart/form-data'>\n\n", "hgPhyloPlace"); cartSaveSession(cart); -char *db = "wuhCor1"; -cgiMakeHiddenVar("db", db); puts("<div class='readableWidth'>"); puts(" <div class='gbControl col-md-12'>"); puts("<div style='font-size: 20px; font-weight: 500; margin-top: 15px; margin-bottom: 10px;'>" - "Place your SARS-CoV-2 sequences in a global phylogenetic tree</div>"); + "Place your sequences in a global phylogenetic tree</div>"); +// If db is not a supported db then switch to the default supported db, and if multiple dbs are +// supported then make a menu so the user can select. +char *label = NULL; +selectDb(&db, &label); printf("<p>Select your FASTA, VCF or list of sequence names/IDs: "); printf("<input type='file' id='%s' name='%s'>", seqFileVar, seqFileVar); printf("</p><p>or paste in sequence names/IDs:<br>\n"); cgiMakeTextArea(pastedIdVar, "", 10, 70); struct treeChoices *treeChoices = loadTreeChoices(db); if (treeChoices) { puts("</p><p>"); printf("Phylogenetic tree version: "); char *phyloPlaceTree = cartOptionalString(cart, "phyloPlaceTree"); cgiMakeDropListWithVals("phyloPlaceTree", treeChoices->descriptions, treeChoices->protobufFiles, treeChoices->count, phyloPlaceTree); } puts("</p><p>"); printf("Number of samples per subtree showing sample placement: "); int subtreeSize = cartUsualInt(cart, "subtreeSize", 50); cgiMakeIntVarWithLimits("subtreeSize", subtreeSize, "Number of samples in subtree showing neighborhood of placement", 5, 10, 5000); puts("</p><p>"); cgiMakeOnClickSubmitButton(CHECK_FILE_OR_PASTE_INPUT_JS(seqFileVar, pastedIdVar), "submit", "Upload"); +char *exampleFile = phyloPlaceDbSettingPath(db, "exampleFile"); +if (isNotEmpty(exampleFile)) + { puts(" "); cgiMakeOnClickSubmitButton("{ loadingImage.run(); return true; }", "exampleButton", "Upload Example File"); + if (sameString(db, "wuhCor1")) + { puts(" "); puts("<a href='https://github.com/russcd/USHER_DEMO/' target=_blank>More example files</a>"); + } + } puts("</p>"); // Add a loading image to reassure people that we're working on it when they upload a big file printf("<div><img id='loadingImg' src='../images/loading.gif' />\n"); printf("<span id='loadingMsg'></span></div>\n"); jsInline("$(document).ready(function() {\n" " loadingImage.init($('#loadingImg'), $('#loadingMsg'), " "'<p style=\"color: red; font-style: italic;\">Uploading and processing your sequences " "may take some time. Please leave this window open while we work on your sequences.</p>');" "});\n"); puts(" </div>"); puts("</div>"); puts("<div class='readableWidth'>"); puts(" <div class='gbControl col-md-12'>"); puts("<h2>More information</h2>"); -puts("<p>Upload your SARS-CoV-2 sequence (FASTA or VCF file) to find the most similar\n" - "complete, high-coverage samples from \n" - "<a href='https://www.gisaid.org/' target='_blank'>GISAID</a>\n" - "or from public sequence databases (" +printf("<p>Upload your %s sequence (FASTA or VCF file) to find the most similar\n" + "complete, high-coverage samples from \n", label); +if (sameString(db, "wuhCor1")) + { + puts("<a href='https://www.gisaid.org/' target='_blank'>GISAID</a>\n" + "or from public sequence databases (INSDC: GenBank/ENA/DDBJ accessed using " "<a href='https://www.ncbi.nlm.nih.gov/labs/virus/vssi/#/virus?SeqType_s=Nucleotide&VirusLineage_ss=SARS-CoV-2,%20taxid:2697049' " - "target=_blank>NCBI Virus / GenBank</a>,\n" + "target=_blank>NCBI Virus</a>,\n" "<a href='https://www.cogconsortium.uk/data/' target=_blank>COG-UK</a> and the\n" "<a href='https://bigd.big.ac.cn/ncov/release_genome' " "target=_blank>China National Center for Bioinformation</a>), " "and your sequence's placement in the phylogenetic tree generated by the\n" "<a href='https://github.com/roblanf/sarscov2phylo' target='_blank'>sarscov2phylo</a>\n" - "pipeline.\n" - "Placement is performed by\n" + "pipeline.\n"); + } +else + { + //#*** TODO get NCBI link from db not hardcoded + puts("public sequence databases (INSDC: GenBank/ENA/DDBJ accessed using " + "<a href='https://www.ncbi.nlm.nih.gov/labs/virus/vssi/#/virus?SeqType_s=Nucleotide&VirusLineage_ss=Monkeypox%20virus%20(monkeypox),%20taxid:10244' " + "target=_blank>NCBI Virus</a>)\n" + "and your sequence's placement in a global phylogenetic tree.\n" + ); + } +puts("Placement is performed by\n" "<a href='https://github.com/yatisht/usher' target=_blank>" "Ultrafast Sample placement on Existing tRee (UShER)</a> " "(<a href='https://www.nature.com/articles/s41588-021-00862-7' target=_blank>" "Turakhia <em>et al.</em></a>). UShER also generates local subtrees to show samples " "in the context of the most closely related sequences. The subtrees can be visualized " "as Genome Browser custom tracks and/or using " "<a href='https://nextstrain.org' target=_blank>Nextstrain</a>'s interactive display " "which supports " "<a href='"NEXTSTRAIN_DRAG_DROP_DOC"' " "target=_blank>drag-and-drop</a> of local metadata that remains on your computer.</p>\n"); +if (sameString(db, "wuhCor1")) + { puts("<p>\n" "GISAID data displayed in the Genome Browser are subject to GISAID's\n" "<a href='https://www.gisaid.org/registration/terms-of-use/' target=_blank>" "Terms and Conditions</a>.\n" "SARS-CoV-2 genome sequences and metadata are available for download from\n" "<a href='https://gisaid.org' target=_blank>GISAID</a> EpiCoV™.\n" "</p>"); puts("<p>\n" "<a href='/covid19.html'>COVID-19 Pandemic Resources at UCSC</a></p>\n"); + } puts("</div>"); puts("</div>"); puts("<div class='readableWidth'>"); puts(" <div class='gbControl col-md-12'>"); puts("<h2>Privacy and sharing</h2>"); puts("<h3>Please do not upload " "<a href='https://en.wikipedia.org/wiki/Protected_health_information#United_States' " "target=_blank>Protected Health Information (PHI)</a>.</h3>\n" "If even virus sequence files must remain local on your computer, then you can try " "<a href='https://shusher.gi.ucsc.edu/' target=_blank>ShUShER</a> " "which runs entirely in your web browser so that no files leave your computer." "</p>\n" "<p>We do not store your information " "(aside from the information necessary to display results)\n" "and will not share it with others unless you choose to share your Genome Browser view.</p>\n" - "<p>In order to enable rapid progress in SARS-CoV-2 research and genomic contact tracing,\n" - "please share your SARS-CoV-2 sequences by submitting them to an " + "<p>In order to enable rapid progress in pandemic research and genomic contact tracing,\n" + "please share your sequences by submitting them to an " "<a href='https://ncbiinsights.ncbi.nlm.nih.gov/2020/08/17/insdc-covid-data-sharing/' " "target=_blank>INSDC</a> member institution\n" "(<a href='https://submit.ncbi.nlm.nih.gov/sarscov2/' target=_blank>NCBI</a>,\n" "<a href='https://www.covid19dataportal.org/submit-data' target=_blank>EMBL-EBI</a>\n" - "or <a href='https://www.ddbj.nig.ac.jp/ddbj/websub.html' target=_blank>DDBJ</a>)\n" - "and <a href='https://www.gisaid.org/' target=_blank>GISAID</a>.\n" - "</p>\n"); + "or <a href='https://www.ddbj.nig.ac.jp/ddbj/websub.html' target=_blank>DDBJ</a>)\n"); +if (sameString(db, "wuhCor1")) + puts("and <a href='https://www.gisaid.org/' target=_blank>GISAID</a>\n"); +puts(".</p>\n"); puts("</div>"); puts(" </div>"); puts("<div class='readableWidth'>"); puts("<div class='gbControl col-md-12'>"); puts("<h2>Tutorial</h2>"); puts("<iframe width='267' height='150' src='https://www.youtube.com/embed/humQ1NyZOUM' " "frameborder='0' allow='accelerometer; autoplay; clipboard-write; encrypted-media; " "gyroscope; picture-in-picture' allowfullscreen></iframe>\n" "<h3><a href='https://www.cdc.gov/amd/pdf/slidesets/ToolkitModule_3.3-508C.pdf' " "target=_blank>Slides for tutorial</a></h3>\n" "<h3><a href='https://www.cdc.gov/amd/training/covid-19-gen-epi-toolkit.html' target=_blank>" "More tutorials from CDC COVID-19 Genomic Epidemiology Toolkit</a></h3>\n" "</p>" ); puts("</div>"); puts("</div>"); puts("</form>"); } static void mainPage(char *db) { // Start web page with new-style header webStartGbNoBanner(cart, db, "UShER: Upload"); +jsInit(); jsIncludeFile("jquery.js", NULL); jsIncludeFile("ajax.js", NULL); newPageStartStuff(); +// Hidden form for reloading page when db select is changed +static char *saveVars[] = { "db" }; +jsCreateHiddenForm(cart, cgiScriptName(), saveVars, ArraySize(saveVars)); + puts("<div class='row'>" " <div class='row gbSectionBannerLarge'>\n" " <div class='col-md-11'>UShER: Ultrafast Sample placement on Existing tRee</div>\n" " <div class='col-md-1'></div>\n" " </div>\n" "</div>\n" "<div class='row'>\n"); if (hgPhyloPlaceEnabled()) { - inputForm(); + inputForm(db); } else { puts(" <div class='gbControl col-md-12'>"); puts(" Sorry, this server is not configured to perform phylogenetic placement."); puts(" </div>"); } puts("</div>\n"); newPageEndStuff(); } static void resultsPage(char *db, struct lineFile *lf) /* QC the user's uploaded sequence(s) or VCF; if input looks valid then run usher * and display results. */ @@ -389,53 +467,54 @@ if (ctFile) { cgiMakeHiddenVar(CT_CUSTOM_TEXT_VAR, ctFile); if (tl.leftLabelWidthChars < 0 || tl.leftLabelWidthChars == leftLabelWidthDefaultChars) cgiMakeHiddenVar(leftLabelWidthVar, leftLabelWidthForLongNames); cgiMakeButton("submit", "view in Genome Browser"); puts(" </div>"); puts("</form>"); } else if (! success) { puts("<p></p>"); puts(" </div>"); puts("</form>"); // Let the user upload something else and try again: - inputForm(); + inputForm(db); } } else { warn("Unable to read your uploaded data - please choose a file and try again, or click the " ""try example" button."); // Let the user try again: puts(" </div>"); puts("</form>"); - inputForm(); + inputForm(db); } puts("</div>\n"); newPageEndStuff(); } static void doMiddle(struct cart *theCart) /* Set up globals and make web page */ { cart = theCart; -char *db = NULL, *genome = NULL, *clade = NULL; -getDbGenomeClade(cart, &db, &genome, &clade, oldVars); +char *db = NULL, *genome = NULL; +// Get the current db from the cart +getDbAndGenome(cart, &db, &genome, oldVars); int timeout = cartUsualInt(cart, "udcTimeout", 300); if (udcCacheTimeout() < timeout) udcSetCacheTimeout(timeout); knetUdcInstall(); measureTiming = cartUsualBoolean(cart, "measureTiming", measureTiming); char *submitLabel = cgiOptionalString("submit"); char *newExampleButton = cgiOptionalString("exampleButton"); if ((submitLabel && sameString(submitLabel, "try example")) || (newExampleButton && sameString(newExampleButton, "Upload Example File"))) { char *exampleFile = phyloPlaceDbSettingPath(db, "exampleFile"); struct lineFile *lf = lineFileOpen(exampleFile, TRUE);