corneliusroemer

// A Newick grammar based on the specification by Gary Olsen: available at
// https://phylipweb.github.io/phylip/newick_doc.html The grammar does _not_ include the optional
// comments as I have not figured out how to include them without adding a lot of `comment?` rules
// to the grammar whick makes it very slow.

grammar Newick;

tree: descendantList rootLabel? branchLength? SEMI;
descendantList: LPAREN subtree (COMMA subtree)* RPAREN;
subtree: (descendantList internalNodeLabel? | leafLabel) branchLength?;

corneliusroemer

import argparse
from pprint import pp

from Bio import Phylo


def get_terminals(clade):
    return frozenset(terminal.name for terminal in clade.get_terminals())

corneliusroemer

{"seqName":"England/MILK-28D52E2/2021","nearestNodes":["B.1.617.2"]}
{"seqName":"USA/CO-CDPHE-2102374275/2021","nearestNodes":["B.1.617.2"]}
{"seqName":"Wales/PHWC-PGY3IH/2021","nearestNodes":["B.1.617.2"]}
{"seqName":"Turkey/HSGM-F70/2021","nearestNodes":["B.1.617.2"]}
{"seqName":"USA/WI-CDC-QDX31282160/2021","nearestNodes":["B.1.617.2"]}
{"seqName":"Canada/ON-SHL-21-02331/2021","nearestNodes":["B.1.617.2"]}
{"seqName":"France/BRE-IPP48628/2021","nearestNodes":["B.1.617.2"]}
{"seqName":"USA/CA-CDC-LC0396661/2021","nearestNodes":["B.1.617.2"]}
{"seqName":"France/OCC-HCL021227635901/2021","nearestNodes":["B.1.617.2"]}
{"seqName":"env/Ukraine/80004/2023","nearestNodes":["NODE_0000637","XBB.1.9.1"]}

corneliusroemer

import argparse
from math import e
from Bio import Entrez
import openai

def search_latest_papers(faculty_name, source="Pubmed", num_papers=5):
    papers_info = []

    if source == "Pubmed":
        Entrez.email = "YOUR_EMAIL@example.com"

corneliusroemer

import argparse
from math import e
from Bio import Entrez
import openai


def search_latest_papers(faculty_name, source="Pubmed", num_papers=5):
    abstracts = []

    if source == "Pubmed":

corneliusroemer

{"version":"v2","meta":{"title":"SARS-CoV-2 phylogeny","updated":"2023-08-08","build_url":"https://github.com/neherlab/nextclade_data_workflows","maintainers":[{"name":"Cornelius Roemer","url":"https://neherlab.org"},{"name":"Richard Neher","url":"https://neherlab.org"}],"display_defaults":{"color_by":"clade_membership","distance_measure":"div","map_triplicate":false,"branch_label":"clade","transmission_lines":false},"genome_annotations":{"E":{"end":26472,"seqid":"defaults/reference_seq.gb","start":26245,"strand":"+","type":"CDS"},"M":{"end":27191,"seqid":"defaults/reference_seq.gb","start":26523,"strand":"+","type":"CDS"},"N":{"end":29533,"seqid":"defaults/reference_seq.gb","start":28274,"strand":"+","type":"CDS"},"ORF1a":{"end":13468,"seqid":"defaults/reference_seq.gb","start":266,"strand":"+","type":"CDS"},"ORF1b":{"end":21555,"seqid":"defaults/reference_seq.gb","start":13468,"strand":"+","type":"CDS"},"ORF3a":{"end":26220,"seqid":"defaults/reference_seq.gb","start":25393,"strand":"+","type":"CDS"},"ORF6":

corneliusroemer

{ "version": "v2", "meta": { "title": "Subtree with Japan\/PG-462868\/2023|EPI_ISL_17243060|2023-02-28 and 51 other uploaded samples", "description": "Dataset generated by [UShER web interface](https:\/\/genome.ucsc.edu\/cgi-bin\/hgPhyloPlace) using the [usher](https:\/\/github.com\/yatisht\/usher\/) program.  If you have metadata you wish to display, you can now drag on a CSV file and it will be added into this view, [see here](https:\/\/docs.nextstrain.org\/projects\/auspice\/en\/latest\/advanced-functionality\/drag-drop-csv-tsv.html) for more info.", "panels": [ "tree", "entropy"] , "display_defaults": { "branch_label": "aa mutations", "color_by": "pango_lineage_usher"} , "colorings": [ { "key": "userOrOld", "title": "Sample type", "type": "categorical", "scale": [ [ "uploaded sample", "#CC0000"] , [ "GISAID and\/or public", "#000000"] ] } , { "key": "gt", "title": "Genotype", "type": "categorical"} , { "key": "pango_lineage", "title": "Pango lineage", "type": "categorical"} , { "key": "Nextstrain_clade", 

corneliusroemer

{"version":"v2","meta":{"title":"SARS-CoV-2 phylogeny","updated":"2023-03-22","build_url":"https://github.com/neherlab/nextclade_data_workflows","maintainers":[{"name":"Cornelius Roemer","url":"https://neherlab.org"},{"name":"Richard Neher","url":"https://neherlab.org"}],"display_defaults":{"color_by":"clade_membership","distance_measure":"div","map_triplicate":false,"branch_label":"clade","transmission_lines":false},"genome_annotations":{"nuc":{"end":29903,"start":1,"strand":"+","seqid":"defaults/reference_seq.gb","type":"source"},"E":{"end":26472,"seqid":"defaults/reference_seq.gb","start":26245,"strand":"+","type":"CDS"},"M":{"end":27191,"seqid":"defaults/reference_seq.gb","start":26523,"strand":"+","type":"CDS"},"N":{"end":29533,"seqid":"defaults/reference_seq.gb","start":28274,"strand":"+","type":"CDS"},"ORF1a":{"end":13468,"seqid":"defaults/reference_seq.gb","start":266,"strand":"+","type":"CDS"},"ORF1b":{"end":21555,"seqid":"defaults/reference_seq.gb","start":13468,"strand":"+","type":"CDS"},"ORF3a":{"

corneliusroemer

{
  "schemaVersion": "1.10.0",
  "placementMaskRanges": [{"begin":0,"end":100},{"begin":29720,"end":30000}],
  "nucMutLabelMap": {
    "174T": [
      "20H"
    ],
    "204T": [
      "20E"
    ],

corneliusroemer

{
  "schemaVersion": "1.10.0",
  "placementMaskRanges": [{"begin":0,"end":100},{"begin":10700,"end":30000}],
  "nucMutLabelMap": {
    "174T": [
      "20H"
    ],
    "204T": [
      "20E"
    ],